I take the point, but I think the author picked a poor analogy. Cooking even an excellent steak is actually not that hard. In fact, I'd argue that it's among the easiest things to master/make at a top level quality at home. Does it require some modicum of attention and understanding? Sure. But starting with a high quality cut, owning a meat-thermometer, and knowing about reverse searing is about all it takes to reliably and easily get a near perfect steak every time.
There are far, far better cooking examples out there.
I'm more of a coffee nerd than a steak aficionado but I've often made the comparison between those two. In both cases, the enthusiast considers their skill in making coffee/cooking steak to be a differentiator and they immerse themselves deeply into the process, skills, and tools. But for both coffee beans and steak the most important factor for a good end product is starting with high quality inputs. In other words, most of the work is being done by the farmers, the processors, and the quality of the raw stock (the cow or the coffee plant). Your job at the very end of that long value adding chain is to not ruin the hard work that others have put into it.
I'm not sure if this analogy holds up in the AI software metaphor that the author of the blog post is making.
Coffee is a good metaphor here too, because bad technique can ruin a good steak just as surely as good coffee. "knows how to use a meat thermometer and reverse sear" describes like 10% of the population or less, same with coffee - there are people who use a bad grinder and a not-regularly-cleaned Mr. Coffee with good beans and get gas station coffee, and there are people out there making solidly B+ class coffee using costco beans, a no name burr grinder, and a french press.
They have a thermometer with multiple sensors in it that finds the core temperature of the food you're cooking automatically, and predicts the rise in temperature over time. So the alarm goes off when it's time to pull and rest, which is absolutely amazing.
They have digital thermometers now that will map out the temperature curve and even predicted when the meat will reach core temp. You really don't need to use them many times before you internalize the behavior.
You'll start to taste the difference once you get that one really-well-made pourover at a place that _lives_ for the love of coffee.
There are some coffee beans that have crazy taste profiles when properly extracted with the right water. I had one that tastes literally like a peach and another that tasted like a really nice tea.
So I can taste differences, but they simply don't matter enough for me. My areopressed coffee with nice fresh locally roasted beans does taste different from a mcdonalds pot. Less harsh, less chemically flavored. But for me those differences don't really matter. They are close enough that I can enjoy both.
But I would say I know I'm also weird in that food temp really doesn't bother me all that much. I enjoy my coffee even as it goes from hot to cold.
A steak for me can be more easily messed up. But then I've not tried instant coffee so maybe I've just never tasted the bottom of the barrel for coffee.
Yeah Mcdonalds is, at this point, acceptably C-grade coffee. You have to go to something like a cafe bustelo in distilled water to get the full contrast.
It’s about experience. If you’re not into classical music, or jazz, or metal, it all sounds the same - how could anybody tell the difference? Only when you put in the work to listen to it a lot, different works, and actually pay attention, you will notice how your sense and understanding develops over time. Suddenly you’re able to make out differences, recognise patterns and repeated segments and structure where before there was none.
Or pick dancing: when you start out, you’re entirely occupied with placing your feet in the right spots at the right time. As you practice, your mind gets freed up and you’re able to make deliberate alterations, embellishments, larger figures. That seems impossible at the start.
No offence, but for some reason some people just can't accept a simple fact: for some people this difference is still not a difference.
I've tried good coffee. At least that's what people told ne on multiple accasions where I was taken to "the right places" in multiple countries. On two sides of the globe (Brazil and Vietnam if we are talking about producers, not to mention countries that import the beans).
Thing is: no matter where it grows, no matter how they prepare it, fermented or not, different frying, different brewing techniques - all the same to me.
I can tell the difference. Some times small, some times very huge. But still for this difference is like a cat piss from some street cat and a cat with a pedigree.
It may have the ideal temperature, ideal acid levels, zinc amounts or whatever - still tastes like cat piss. The fact that it has a pedigree does not make it better.
Tea is whole different story for me though. But I do accepts that people are fine with whatever they love in their "english breakfast tea" or a nameless teabag.
That likely means that you've only had "bad" coffee. You're one of today's lucky 10,000 - go treat yourself to a decent cup of coffee it tastes completely different. The difference between a good steak and a bad steak is actually a good comparison.
Sometimes it's a matter of not having taste for it. If you give me multiple bottles of alcohol drinks, I might be able to distinguish them, but I have no clue at all which ones are supposed to be good. To me it's all just alcohol. Same with coffee - but not the same for steak.
I thought that too, honestly, before I had truly great coffee. I honestly didn't like coffee at all, and then a coworker brought in some coffee he'd roasted and prepared himself and it was lifechangingly good. I also made myself incredibly sick with the enormous amount of caffeine, it's been very rare that I've had to go home from work in the middle of the day but that was one.
Pickling a cucumber sure has a lot of resemblance - a great cucumber is good fresh and great pickled, but a bad pickling brine can ruin even a great cucumber.
I don't know if I agree with the take I've written below, but I think you can make the analogy work:
It has been argued, even in the before-AI-times, that most professional software is just gluing together open source tools and libraries. Your job isn't to write software, it's to set up redis and postgres and a programming language and a react frontend or whatever in a way that lets them do what they're there to do. Most modern programming is just configuration of pre-existing tools. It doesn't really matter how good your code is as long as it isn't terrible, Real Programmers already did 95% of the necessary work before you decided to build your little app.
But the point is that no matter how well you cook the steak you're not creating quality, you're just getting the most out of what you have. The cow and the farmer created that quality. I believe we lose sight of that because so much of the value chain is not visible to us and we hyper fixate on the fun part, becoming a better barista/chef.
Yes. It is about learning all of these things. Knowing what kind of cut you have, and how to treat it to best effect.
A very tough gristly cut can be improved by tenderizing. Going low-and-slow/sous vide to melt the connective tissue. Even grinding it up to make a hamburger steak / loco moco.
Once you know all of those things, choosing the right approach and executing it is not that hard. The steps themselves are simple.
How many people who want a good steak know how to identify a high quality cut? Do they even know where to buy a high quality cut? What the different cuts are and whether they want rump, sirloin or fillet? How many kitchens have a meat thermometer? How many home cooks know how to use it and what the right temperature should be? How many home cooks know about reverse searing? Or timing? Or resting? Or seasoning?
Is it all learnable? Sure. But even something as simple as steak has nuance that needs to be learned, equipment deployed to go from "pretty good" to "great", experimentation and trial and error.
Steak is the simplest thing to choose to get started. It's also the hardest to get right.
Building an app is the simplest thing to choose to get started. It's also the hardest to get right.
So, honestly, the whole metaphor stands up pretty well for me, specifically because of your response.
My point is that steak is very, very, very far from the hardest to get right. I'm not going to contest your argument about how hard it is to get right, I'm going to argue that innumerable other dishes are vastly harder. Because the quality ingredients are harder to find, because they involve far more steps, each of which is harder to judge, because often good recipes or instructions simply don't exist, etc. etc. etc. Every single difficulty that you imagine exists to make a truly great steak exists 100 fold for other dishes.
In my opinion, steak has gotten the reputation that it has because it's so easy to get top tier restaurant quality at home (relative to cooking more generally). It is easy enough that, with a little care, almost any one can master it if they decide to, and therefore lots of people try and end up caring, and discussing, and sharing tips, etc.
There are a whole host of dishes where true mastery, to restaurant level, is borderline impossible for a home chef, and so most people never even try.
I don't disagree that reliably getting the last 5% of quality out of a steak is non-trivial, but the idea that, relative to the entire world of cooking, it's particularly difficult is laughably incorrect.
Go to costco, buy prime filet. Put it in a ziplock in a sous vide at 130 for an hour. Drop it in corn oil at 425 for 30 seconds. Add some butter and salt.
That simple process, suggested by Nathan Myhrvold (former MSFT cto) in the modernist cuisine, produces a filet better than most restaurants.
Since we're deep enough in this thread where we're talking about steak rather than AI, I'd like to lean into that.
I used to sous-vide a lot. It was certainly great to get perfectly-done steak. but I still didn't love the result and it took me a while to figure out.
Specifically: sous-viding at 130 doesn't reliably render fat. So it's great for a lean filet mignon like your suggesting, but kept giving me rubbery fat in anything with more fat or significant marbling. I eventually switched away from sous-vide (I usually either grill or reverse sear, depending on my effort/time availability) for most cuts these days.
For a filet for 1 hour, you can sort of do what you want. I add a tiny bit of olive oil and salt, but not much. If you start going longer times with the sous vide, you can get different reactions so you have to be more careful.
That "simple process" requires equipment most people don't own or know how to operate, an oil most people don't keep in their cupboards, and is suggested in a book most people haven't heard of by a tech CTO they would not recognise the name of.
Honestly this is like the Dropbox comment all over again. Everyone is missing the point, it's glorious. :)
Most people cooking steaks don't care about the optimal way to cook steaks, or want to find out.
Most people building software today don't care about the optimal way to build software, or want to find out.
That's TFA. That's the point. That's what we're discussing. :D
I used to sous vide steak in college with just some hot water and my roommate's six-pack cooler. Water should be no more than 5F warmer than the cook temp. For a medium rare steak, cook temp is 129F so get close that. Salt the steak no more than an hour before you start. Steak goes into a Ziploc bag, submerge it up to the zipper to let water pressure force out the air and then seal it shut. Close the lid and come back in an hour, two hours if it's a thicker steak. Take it out of the bag, blot it with a paper towel, apply some steak seasoning, and toss it on a pan at the hottest temperature you can make it. Wait 3 minutes and flip, wait another 3 minutes and serve immediately. Toss a pat of butter on top while in the pan if you're a hedonist.
Only prerequisites are a cheap kitchen thermometer, a ziploc gallon bag, and a cheap cooler. Everything is available at your typical supermarket or even goodwill. This is a very hands off cooking method and requires no experimentation, just follow the recipe and you get a perfect steak every time.
The key is that a cooler holds hot things just as well as cold things but without a circulator and heater to keep the water warm, you're at the mercy of the cooler's R-value. The water will eventually get too cold to effectively cook the meat so you're limited in the length of time you can sous vide. Doing a 24 hour chuck roast is simply not going to work unless you're refilling the cooler every couple hours.
Corn oil is cheap and available at costco. A sous vide is a fancy french name for a $150 device from target that keeps water at a specific temperature...
This "simple process" is much, much simpler than most meals one cooks at home. It's not much harder than microwaving something.
You are missing the point. With a small change in the way you do something, the really good thing can be made.
Reminds me of recipes like brining a turkey in Dr Pepper... sure it may taste fine, but I'd rather not eat food prepared in plastic bags with processed oil.
All software to an extent is a model map of some real world phenomena. The limitations of which can actually be advantageous, a good example is why games with poor graphics can be so immersive.
Sure there are a few things that knowledge and skill can do to make a steak better or worse. But the entire range from worst to best is just tiny. The best steak in the world is not a steak moon shot, it's just barely any different from what anyone can do after barely any instruction.
It's a tiny thing to become about as good as it's possible to get.
I wonder if something is lost to translation. Cooking a stake could be the simple act of putting the steak in the oven with basically zero seasoning with a thermometer and then wait until medium rare with some optional searing. It could also mean trim the meat, cut into slices, seasoning, and frying/grilling them to each persons personal taste, which include well-done.
Perfect grilled steak, especially well-done, is something I find very few people have the experience or skills to do well, especially over a coal grill where the heat won't be constant and the person also need to first have good experience in making a good and even coal bed.
Just to add my favorite fool-proof method to the others already listed here.
A meat probe and the 30-second flip technique is a stupid-easy way to a perfect steak every time. Just make sure you've got great ventilation. Also salt the meat 1 hour before.
It might take more than 10 years, but robots will almost certainly master cooking single pieces of meat.
It's substantially harder to reach self driving than it is to cook a "good" steak.
Sure, for probably quite a long time you'll be able to find someone who can cook a way better steak than the best robot, especially because taste is extremely subjective...
You can debate the horizon and the extent, but changes ARE coming.
Change is the only thing you can count on besides death.
How many analogies age more than 10 years anyways? They are contextual. It will be a while before a robot flips steaks cheaper (and better) than a human line cook.
Is it moving the goalposts to ask Waymo to demonstrate a drive across America and not in well mapped cities with only lower speeds? How did we define "self driving"?
I'm still working on collecting the drive to all states badge. I've only accomplished half. I calculated my total driving since getting a license recently and its approx. ~360k miles or around 17k miles a year.
If Waymo can eventually handle rural life in the US I'd be seriously impressed.
Agreed. I'm an avid cook. Sous vide circulators are cheap now, and so are meat thermometers. Doing a steak well is one of the easiest cooking tasks, to the extent that I won't bother paying top dollar in a restaurant because it's so easy to do restaurant quality at home. I could teach someone to cook a restaurant quality steak in a couple hours and a few practice runs.
I've made steak probably 100 times in the past decade (sometimes with a sous vide) and I don't think I've ever made one I would say is great. Certainly not as good as what I can get at a good restaurant.
I think what my home steaks are really missing is the crust from salamander broiler. I don't have anything at home that can exceed 1000 F / 550 C.
There's so much written about how to cook a steak well, and I've tried a lot of exotic techniques (I have gas and charcoal grills, tried cooking straight on a chimney starter, have gas torches inside and outside). Most of what really matters is that it's dry, the pan is heavy and hot, you use enough oil to get good heat transfer.
If you get those parts right (and control the inside temp via sous vide or reverse sear) most of the other stuff is pretty marginal in how much it helps. Even all of the extra herbs, and butter basting and everything -- relatively insignificant compared to just getting the salt-level right.
Salamander broiler would be incredible to have at home. We get decent results from using a cast iron pan thats as hot as we can get it. At most we can sear 2 steaks as it tends to cool off pretty quick even under high heat. Got any better alternatives that a home cook could use?
Are you just heating the bare cast iron to crazy hot temperatures and then putting the steak in?
I've thought about trying that, but I'd need to first strip the seasoning off the pan to avoid all kinds of nasty flavors. The second pan makes sense...
Reverse sear over sous vide -- moisture is the enemy of crust.
Cheap knock off a Big Green Egg with hardwood charcoal can get to 650°F or so pretty quick, and does an amazing job of finishing up after a reverse sear.
I feel like I am better off with cast iron or stainless steel compared to slightly bigger torch. Stainless steel allowing some deglazing for sauce. Torch is still too small area and burns more so sears.
Agreed. With steak, the main point is to let the quality of the meat speak for itself. Too much input from the cook and you just get in the way of this.
How can I make a really great steak myself? What to look out for, what to do. Give me detailed instructions.
I got a really good answer (including - cut, temperature and reverse searing), so I probably won't even need understanding. (Oh - and yes, I can cook - the answer mirrors what I already know).
Yes, it was a very, very poor analogy, because cooking a steak is a well known and well documented thing where LLMS produce good output. They fail when you get off the beaten path and try to do something that's not "mainstream".
While smoked turkey (from Reese bros) is easily the best poultry I've ever had, I certainly would not call poultry the best place to start learning smoking.
In general I'd recommend starting with the most forgiving meat available to you (IMO pork butt or shoulder are great options) as even when you don't hit the ideal cook you often still get enjoyable results. Repeating the process enough times to learn how to dial in a cook is much easier when the misses don't suck.
Funny I tried the same and it advised a using searing/smoking hot pan. Which is poor advice unless you enjoy the bitter acrid taste of burnt meat. The Maillard reaction doesn’t require those sorts of temperatures (under 180C is hot enough) - especially as the surface of the steak will dry from the oven phase of the reverse sear. “You were right to pull me up on that…”
AI has proved several times they can prove some complex math things that humans have been trying for decades so solve. They have not failed and have clearly done complex things humans cannot. Since humans have not done it before you cannot call it on the beaten path.
> I take the point, but I think the author picked a poor analogy.
He may have missed the point.
AI-generated code does not seem to have parsimony, for one. Generating code is cheap, so there's no drive to condense it. LLMs are not yet good at formulating abstractions to make code smaller, leading to code bloat. It's the same phenomenon that appears when humans modify code they only partially understand by adding special cases to general code.
There are complaints about "elegance", but it's more useful to talk about parsimony. Code bloat is measurable. Elegance is hard to assess.
The problem is that not everyone can afford/wants to spend on a high quality cut. A good cook can make a cheap cut of meat taste great. There's no single right way of cooking a steak and that can include spices/herbs/marinades/oils/etc.
It is a great analogy, as the vast majority of software tasks are CRUD or glue logic. At this point, relatively narrowly-defined tasks are a solved problem with frontier models. It is only more advanced tasks with broad scope and large context that frontier models have more limited results.
The first, and by far most important is that unlike ground meat, your bacterial concerns are almost entirely on the surface. Unless something very very very wrong has occurred in the processing and transport chain, bacteria will not have penetrated into the meat. And if it's gotten bad enough that they have, you will almost certainly be able to see and smell it. And the surface is cooked to a temperature far in excess of that needed to kill bacteria.
And even for the interior, rare is not the same as raw. A completely rare steak should be cooked to between 120 and 130 (if you want it "blue" you might go down to 120). That's enough to get a decent reduction in bacterial count on it's own, although admittedly many techniques will not hold it long enough to get the number of log reductions that food safety guidelines recommend, in the case that you somehow did have a large bacterial infection in the interior.
So the answer to your question is that if you are buying your meat from an even halfway reputable source, the only place that bacteria might exist is on the surface, where it has been cut by tools and handled by the butcher/you and in general come into contact with the world. Those bacteria will be 100% killed by the cooking process. The odds of getting sick from a rare steak are low enough that if you are concerned about that, there are innumerable other things in your life that you should be more concerned by.
This is why rare steak is ok, but rare ground beef hamburgers are not. Although less cooked through burgers are sometimes served in restaurants that grind their beef fresh right before cooking
It also doesn't help that the quality of ground beef in the US is much poorer than other countries, so those who may regularly eat rare ground beef internationally would experience a level of shock if eating it in the US.
I mean, rare ground beef hamburgers are pretty much fine too. They have a higher risk of contamination, yes, but that does not make it a particularly high risk. I eat burgers, made from pre-ground beef, cooked rare all the time and I have yet to get sick from it even once.
Great answer. Just to add, this is with red meat where E. Coli is the main concern. That exists in the gut of the animal and that bacteria can travel to the surface of the muscles during the butchering process. Poultry on the other hand is mainly concerned with Salmonella which does live in the muscles of the animal. That's why people aren't ordering rare chicken breast.
Also, rare chicken breast is apparently pretty unpleasant to eat. Kenji Lopez-Alt did an article where he discussed how you can use sous vide to pasteurize chicken while still keeping it at a low temp, but the result (while perfectly safe to eat) was not enjoyable.
Am I the only one who thinks the "reverse sear" is like cheating? It takes out all of the skill when you move from the pan to the oven, and I lose all of the fun in learning to be chef.
It's one of those things that you have to experiment for yourself I think. You'll get tons of advice, I already saw sous vide, reverse sear, 30 second flip, etc. My favorite? I just get a cast iron as hot as I can on the stovetop and flip it once, then at the end quickly sear the edges using tongs.
To get a perfect steak with nice X-shaped sear marks on both sides:
1. Turn grill on all the way. Close lid. Wait for grill's gauge to reach >400 F (>200C).
2. Put steak on grill. Close lid. Wait 2 minutes.
3. Rotate steak 90 degrees, placing on a fresh section of the grill (not where it just was). Close lid. Wait 2 minutes.
4. Flip steak over, placing on a fresh section of the grill (not where it just was). Close lid. Wait 2 minutes.
5. Rotate steak 90 degrees, placing on a fresh section of the grill (not where it just was). Close lid. Wait 2 minutes.
It's been 8 minutes of grilling. Steak is about done. Stab it with a meat thermometer into the middle part, it should read 350 F (or 176 C) to be medium done (pink in middle).
Tips:
a) Optional but recommended: rub steak with olive oil and steak seasoning on both sides, including the edge.
b) If steak is thicker: add a little more time. If thinner, use less time.
c) If steak is frozen (buy good cheap steaks and freeze to last a long time): add a little time.
d) To cheat, put a little dab of butter on top for the last segment of each side.
e) Leave the steak out for 15-20 minutes to get to room temp before making it.
f) Let it sit on a plate for 8 minutes before serving. It's still cooking, and the juices settle in. You can "tent" foil over it if you want.
g) Don't cut the steak when checking it. "Fast" and accurate meat thermometers are cheap now.
This is for grilling. The stove, broiler, and cast iron skillet instructions are different.
8minutes, that's just pure waste; even medium rare (which is hardly tolerable) is 56-57C. Also cooking the steak without any salt is travesty (that should not be an optional step)
I do not appreciate when authors use the royal "we" to speak for all software engineers when admitting to low quality-control standards.
I suspect that it is an attempt to broach an uncomfortable topic through vulnerable self-disclosure, but we need to be serious about admitting when there is a problem somewhere.
"Bugs" are not any more cute or fuzzy or entertaining or harmless than the engine "gremlins" that haunted the aviation industry back in the day.
I don't know how many accidents had to happen before the airplane people got serious, but software people are overdue for a similar reckoning.
I just want to point out this is not a royal "we". It's a regular "we". Royal "we" is when you say "we" but mean a singular person, yourself. Here, the author does actually mean multiple people beyond themselves - like you pointed out, they claim to speak for all software engineers.
No, that is exactly GP's complaint. Their whole point is that OP has written things like "we throw everything at a model and hope it gives us what we imagined, without ever having to understand how any of it actually works" when the honest statement would be "I throw everything at a model and hope it gives me what I imagined, without ever having to understand how any of it actually works".
> To build good software with AI, you still have to understand software.
> You need to know what you’re actually asking for, how to judge what comes back, and when the machine is just confidently serving you charcoal.
It's surprising to me this even needs to be written down. Like, obviously still use your brain and think about what the computer is doing. Did you guys think that LLMs were magic?
Article aside, I generally always use the "we" pronoun at work when writing prose. Saying "I" feels too adversarial when talking about negative effects, or too self-aggrandizing when talking about positive effects. For example: "We discovered a bug shipped at the end of the merge window, so we will restart validation with the rollback applied." I think in school they say the business-safe way to write is instead with the passive voice but I just can't do it.
"We all could have done more.": said by every politician. Which is easier: writing good code, or winning a political battle over crappy code? I mean, if you're not capable of coding, and creating an appropriate architecture for the the domain: which course of action gives you the most control over risk management? So yeah most fights about software are about politics and crappy code.
Saying this to set expectations based on having a business license since 1984.
This premise seems incorrect. SWEs may aspire to produce perfect steaks as a matter of pride, but demand asks for mass produced steaks that trade off against other factor like time and cost.
There is a reason github isn't full of nasa lunar lander quality this can't crash code...
I hope and believe that some great companies and teams will use LLMs to build higher quality software.
We built Electron because writing UIs using native desktop frameworks is tough. Is it still tough? Surely LLMs make it easier to use and so we will end up with faster and more native feeling applications.
What about having a couple of ideas of what might make a feature feel good? Well now you can make multiple prototypes fast and pick the best one. Your users get the best one.
That was such a weird move for them to make. If you have AIs that are good at building software, then use them to build great software! Instead they choose lowest-common denominator solutions that are okay everywhere but not great anywhere.
Cross-platform widget engines (Tk, GTK, Qt, wxWidgets, etc…) long predate Electron and have always been robust, stable, well documented, and easy to use. It's not creating a single application for multiple native desktop environments that was difficult, it's writing a single application as both a web interface and a native desktop environment on any platform that's difficult.
Because CSS is constantly changing, inconsistently implemented, and difficult to use, it can't easily be targeted by widget engines (although Qt has tried, and maybe others) and web browsers already exist, so webview environments like Electron are the quickest way to get a cross-platform installable application that can also be hosted on a web server. I wouldn't really call it native though.
It will be a push and pull and I'm curious where we'll be in a couple of years.
I can fly at 100mph if I let AI run loose and with a bit of steering I can get it to output what I'm looking for and generally pass verification and tests.
If I care about the code though, and I want to keep it maintainable, the amount of time and tokens I need to spend correcting and iterating on the output quickly eats through much of the initial time I saved, to the point where I'm unsure if I'm actually saving much time at the end of the process.
With hobby projects I lean on quality more, and the async nature of AI also makes this much easier to make progress without needing my full attention to do so.
In the corporate world, there's both the pressure to accelerate with AI, but also maintain code and product quality. The dials of one way or the other are more obvious now, but I don't believe it's possible to do both with the current models and harnesses without exponential cost.
What is interesting though, is I'm now leaning towards faster models rather than smarter ones.
Intelligence lets me bite off larger chunks of work at once, and trust the model to behave without having to watch it intensely, but doesn't seem to drive down the number of iterations required to hit my desired quality.
Faster models means the iterations I'll have to go through regardless will complete much faster and gets me closer to a proper flow state. Models will keep improving, but maybe we're getting near "smart enough" and the race will pivot to performance > intelligence.
I also spend a lot of time at work trying to get the code into a shape I deem acceptable to open a MR for the team, without looking incompetent.
At home I stopped reviewing the code for my mobile app, and I do not feel this has affected maintainability.
With 150k LOC, the AI still performs changes as easily as it did three months ago, when I started the project.
If anything, the new models made it easier.
When I read all those posts from people advocating a very hands-on, understand each line approach, I speculate many come from a position like me at work or have never seriously tried 'vibecoding' a large project from scratch with SOTA models.
People routinely post that things I work on for 30 hours/week are impossible.
> We built Electron because writing UIs using native desktop frameworks is tough.
No, we built Electron because there are a huge number of people who are too lazy to learn anything outside of HTML/etc for creating a UI. It's not actually hard to make a UI using the native desktop frameworks.
>We built Electron because writing UIs using native desktop frameworks is tough. Is it still tough? Surely LLMs make it easier to use and so we will end up with faster and more native feeling applications.
We built Electron because web devs were a dime a dozen. It was an economic decision, not a technical one.
You would think so. But it seems all that everyone is obsessed with is increasing the velocity of enshitification in the hopes of becoming the next papa Elon
The best framing I've seen so far—and boy have a lot of people tried to frame the capabilities of AI recently—is that AI shines for people who use it to extend their thinking, and fails for those who use it to replace their thinking.
The latest frontier model developments have chipped away at that latter category. But ultimately the ceiling for that is mediocrity. Yes, if you imagine the most average common thing that has been done a million times, AI can absolutely give a bespoke version of it that matches your words of choice. For the moment it's still an impressive parlor trick to a lot of people, but the novelty wears off fast, and then you realize there's very little value in that. Doesn't mean everyone has to learn to code to use AI effectively, but it does mean they need to be curious and engaged or they'll just become part of the increasingly irrelevant slop machine.
Interesting, the NYT had a tip on flipping every 90s, but it was with a cast-iron pan and not a "cold sear". That was too smokey for our weak kitchen exhaust, so it was the last time I tried flipping multiple times.
But Lam Lam's cold sear video doesn't seem to make that much smoke on a non-stick, so I'll give this a shot again.
(Personally the reverse sear has been my idiotproof go-to.)
It would certainly help if it could taste what it produced. Or see, with its eyes, how the result is changing as its process changes. Or learn, from first principles, what is happening at each step and make deductions about how they will affect the result.
All it can do is randomly shuffle around the steps and ask if the customer likes it. If they don't, shuffle them around again and hope it's better this time.
Don’t forget the moment where you realize that instead of cooking a steak the agent went for its own invented requirements and spent all your quota preparing everything to make the perfect ramen from scratch (or at least that’s what it tells you, when you ask it to double check what has been done it comes back and tell you it lied and in fact mixed the flour with the soup without making the noodles, with a small note telling you to not forget to prepare the noodles when you go to production)
We have this idea of AI software development where a human watches over its shoulder and shouts out things like "use uuidv7 for the id since timing is important" or "use a sum type here to make that bad state unrepresentable".
But the thing is that most of this can be encoded in a markdown file for agents to read if it doesn't already come out of the box in the next round of sota models. And funnily enough as agents get smarter, you risk being overprescriptive where you hamstring the agent from making a pivot that would have led to better engineering.
The future is pretty clear to me at this point that we won't need software engineers looking over the shoulder and instead it's just a "user with taste" asking for revisions.
Have you ever been really deep in a thread with an agent on a technical deep dive? The trade-offs conversations would not make sense to someone that is not able to decode the real decisions they’re making.
You can dumb down the agent, but the reason domain language matters is because of the precision of the word sometimes - not to be a gatekeeper.
Anyway - user with taste that understands all the elements of their decisions would be my modifier :)
Long ago I added an instruction to my global AGENTS.md that every time the agent asks me something, make it multiple choice that ranks the top options, give a quick trade-off summary of each one, recommend one, and justify it.
Over the last year, I've noticed that the recommended option for technical questions has become one that is always sufficient to the point that I'm once again in a tastemaker position, not in the "here's a better idea btw" position.
And if that's true, then my participation in the intermediate steps isn't strictly necessary to get a well reasoned position that's only directionally wrong.
I was once impressed by a very crappy cut, a lengthy marinade with a lot of vinegar, a rins and a extremely sweet saus. The meat was very red and was as soft as bread.
But so much of the AI capability discussion can be avoided once everyone understands that a) There is some multivariate supply and demand market for software and b) AI is shifting the supply curve of software.
The shift is uneven, not perfectly monotonic, and might actually result in shifts in the demand curve (as e.g. 10x seniors get more valuable).
This insight makes it easier to see the forest despite the trees.
I like the comparrison of the AI coding agent to the steak machine. But it is not just about machines scaling better - the laundromat can wash clothes better than a human would. Similarily the AI agent's performance can EXCEED one of a human engineer for simple, well-defined tasks (eg. go over all of the subpages and check that the latest push didn't break something).
Hiring AI cooks is not a problem imo, as long as the human glances over everything that they serve.
In the end, unless we are talking about art, in most areas I probably prefer something that follows some specific set of requirements and recipes, rather than have some sort of indescribable 'magic'. In that sense when it comes to coding AI gives you something that can follow the requirements pretty well and thus satisfies me in many circumstances. (but there are cases where human input is extremely important still)
Cooking a genuinely good steak also requires almost no skill. I can give you a chicken scratch note with 3 bullet points and it will come out nearly perfectly.
I was going to say the same thing. 2.5m per side+rotation plus another 2.5 plus or minus 20s depending on heat
But I actually got something out of this -- which is that maybe before now, humans had to soak up the complexity of where value was made in software. And you saw this in the highest and best paid engineers in science and platform teams and basically that you had to invest in brains where you thought the value in your product was.
Without humans managing complexity, you have no idea where you should be investing once the product grows out of the concept stage
Slight disagree, as many will heat their pan too high and burn the steak before it cooks through. Or maybe that was just me when I first learned how to cook steak the right way. ("Medium"/"medium-high" doesn't necessarily correlate to the position of the knob on the burner or whether the pan feels medium/medium-high)
Well it's not like most companies can pay for freshly made software; most software today is made for a wide audience, bought frozen and reheated in the microwave.
Maybe freshly prepared software made at home in a less-than-perfect manner for a tiny target audience is still a mouthwatering dish in comparison.
Using Claude Code is like asking questions in online forums, if anyone remember that, we still have sites like Stack overflow. The difference now is that those asking the questions are actually paying for the answers.
I prefer the other end of the curve (if that makes sense) where something is time consuming and hard to make at home YET you can get it incredibly cheap comparatively and in good quality in a restuarant. Ramen is the prime example of that.
Ramen may still be somewhat of a novelty where I live, and I would not call it cheap. Though now that I put that into words, a decent burger anywhere is about the same price..
It is compared to the cost of making an authentic ramen broth at home at the scale of feeding 2-4 people. Pho has a similar property. I've made homemade pho broth once and it was very hard work, fairly costly, and it resulted in something that was pretty good, but not comparable in quality or price to something I could have bought in a restaurant down the street. (admittedly typically also cheaper that ramen)
I'm tired of articles which try to make us not feel bad about being eclipsed by AI. We've got to change from the mindset of an artisan--who stakes his being on mastering skills and acquiring knowledge--to that of executives.
Figure out what you want to DO. What you really want to do, not what you do just to pay the bills. Once you figure that out, use AI as an executive uses the best employees he can buy.
Given the state of software today, I'd be happy with steak that was competently cooked. I love software development, but most of it is not high art, nor high performance, nor free of bugs.
I like a good steak but setting off the fire alarms (plural) is a big risk when I cook them in my apartment :-). Scorching hot cast iron pan plus meat means some smoke is unavoidable. And that's the proper way to do it.
The problem with many business leaders is that they don't necessarily understand the businesses they lead very well. That's common for CEOs running a tech company without a tech background. There's a long history of business leaders with MBAs throwing away the baby with the bathwater when they get a bit too enthusiastic cutting cost. It's why Boeing is such a mess. It's why Intel is no longer considered a leader in innovative chips. The likes of Nvidia and Apple are running circles around them. It's why IBM is now a consulting company that hasn't really produced any tech product worth mentioning this century. The only time they are in the news is when they do another layoff round to make share holders happy. They are incapable of innovating because the people that did that have long since left.
In the hands of somebody competent, you can do great things with AI. But if you can't tell good from bad technically, you won't know that you are creating a big stinky mess until it's far too late.
On a positive note, there will be plenty of work for competent techies to come in and straighten things out on a freelance or consultant basis (I'm available for that sort of thing but not cheap). With the help of AI of course. No need to go back to doing things manually. I think a lot of companies will realize they need help in the next years to speed things up, to improve quality, etc. They'll be very opinionated about what they want and used to getting what they want quickly.
So use AI like a sous-vide: have it tackle the fiddly, difficult, unglamorous parts using strict controls and then do the parts that require taste yourself.
It's a level of badly written that makes me think it could be a human, or it might just be an LLM that writes differently from a typical ChatGPT article.
About five years ago I flash-fried a frozen steak.
Now I only eat steaks grilled from frozen. Get the grill nice and hot (450F at least) and throw the frozen steak on the grill. 5-10 minutes each side usually gets it nice and crispy on the outside while rare and tender in the middle.
You can still get quality services and products. It's still possible to get a well cooked steak, but they are going to get harder to find and vastly more expensive and many people will be priced out of them entirely and will have to settle for the slop (which will also only get more expensive even as quality gets worse).
Between "AI" incompetence and companies not caring about doing anything above the bare minimum it's making our lives worse and more obnoxious. AI is accelerating the enshittification of everything and it goes way beyond software. It'll now be your responsibility to spend your free time on the phone talking to a company's AI chatbot "customer service" trying to fix whatever problems their use of AI will inevitably cause you.
When we envisioned our future we thought everything would get better. All high tech and fancy, but it's looking like what we'll get is expensive yet barely functional and annoying while we collectively just get poorer and dumber.
I think most people on this board would agree with the thesis. However the real problem is when the owner of the restaurant looks at profit and loss statements and decides to keep less chefs on the payroll because customers are willing to pay for (just) edible steak.
The reduced expectations of steak, the desire for the perfect steak, are all fading to the background because "just passing satisfactorily" is better for business.
I love a good medium-rare prime steak, but I also used to live near a restaurant called Best Steak House, which was anything but. However, they delivered a passable steak/steak sandwich that was worth what you paid for it. And considering that they were in business for decades, probably most people felt the same way.
A great steak is an occasional luxury; a "just edible" one is an everyday meal.
How are you still blaming this on “the government”? Is your thesis that the business world would be delivering high quality products and experiences if only those damned bureaucrats would get out of the way? What possible evidence do you have left for that hypothesis at this point?
EDIT: this was a misread on my part, retracted, carry on.
He is saying "if we apply (or are applying) the same business mentality to government, we will get literally the worst government we are willing to tolerate based on money"
It's not blaming. It's an explanation for probably why things can go wrong.
He isn't blaming the software problem on the government. He is saying that if we apply this reasoning to everything in our lives, then we have a race to the bottom in other facets, of which one is government.
Not always. There are diminishing returns at the top end of the quality spectrum. I tend to go with the cheaper item even if I can technically afford the more expensive one if their quality is close enough.
The thesis is covering the flat scenarios mostly. I think the scenario where the job is done so poorly that the steak has to be thrown away needs to be included as well. At times the pans would get damaged. And eventually the business is in debt and needs to be closed
I can go out today and pay $50 for a thick cut ribeye at Ruth's Chris that has been visually inspected for marbling then hard seared in a special high temp over to temp by someone who knows wtf they're doing. I can also go to Denny's and pay $20 for a t-bone cooked to within shouting distance of what I asked for on the same griddle they use for the pancakes that doesn't get the sear I'm looking for.
Point being, you're absolutely right that the goal for a company selling a product is to be barely satisfactory but they don't actually get to decide what's satisfactory.
The main problem with this entire article is that sometimes, if not most times, you don't need great steak. Sometimes you just want to satisfy your hunger and if the steak is lower quality meat or if it's overcooked, you don't care. All you want is to not be hungry.
This is what vibe coding satisfies and frankly what most people care about. 85-90% of human written code is garbage anyway. There's nothing sacred about human-written code.
Most customers don't care about the quality of code as long as the end product works. The great thing about vibe coding is that if something doesn't work, you just ask it to change it and within a minute you have the change. You don't have to send off a request to an offshore contracting team, and go back and forth over what it should be, haggle over hours, and then have it come back with some deficiencies because they didn't follow the agreed-upon spec.
There is a lot of truth here, but it clearly isn't a welcome one.
From my own perspective I can look back on decades of software I've written (some including code generation far before LLMs), and marvel at the quality and creativity, and lack thereof, from one piece to another.
I never meant to create awful software, but I did (and still sometimes do). With LMM code generation (vs bespoke tooling, T4, XSLT, and the like) the game of chance is part of the fun, harnessing a powerful tool that wants to run out of control on a whim.
Looking back on a couple years of LLM assisted work I see the same mix as the decades before: some great (when I managed to keep the beast restrained) some awful (when I didn't, knowingly or not).
I don't see how things are much different with this tool than others as far as my work product goes. There is a bit more of it, but the excess isn't great stuff (that quantity remains roughly consistent over the years).
I was going to say something similar. In particular, I do a lot of prototyping, AI has allowed me to make prototypes an order of magnitude faster with substantially higher quality. Different types of software engineering have very different goals and standards. AI does cook the perfect medium rare steak every time, but it's able to actually cook steaks in the time it used to take me to like microwave a burrito.
I suspect that there is a lifetime of knowledge and experience wrapped up in that statement and I agree wholeheartedly.
I love writing software and I've been doing it for decades. But I never forget that at the end, there's someone who's paid for it, and needs it to do a job that can't/won't be done manually. And that's why it exists. Not to satisfy my desire to express myself in code, not to allow me to create some golden tower of perfect architecture, but to do a job. Most likely a boring job in the service of increased profitability for some company.
And in the end, the only thing that matters to the customer is if it does that job well enough to be useful. I think -- hell, I know -- that many developers deliberately ignore this critical point.
Programers are high end chef's whos products are expensive.
LLM / AI tooling is going to do a much better job of delivering on the promise of AppleScript, vb script, IFTT, and every failed drag and drop coding tool that got sold to businesses over the years.
There are going to be gains in large software, from professionals - but the real gains are going to come from all the software that CAN get created that never was before.
There is an XKCD comic "Is It Worth the Time?" (link: https://xkcd.com/1205/ )- a matrix of the cost benefit of automating something. LLM's / AI tooling, generating traditional code, makes the answer "yes" in almost every case now.
Software slop has existed for as long as software has existed. AI is now a better coder than the median. By the end of next year it'll be better than the best human coders.
Steaks require very few skills. Buy 10 steaks, some oil, a pan, and a heat source. By the time you've cooked the 10th steak you will be able to cook a decent steak.
Software requires a massive amount of skill. You can't say you can build "consistently good software" after writing your 10th program. Most software engineers really aren't great at judging what makes good software. So humans aren't a great solution to AI's lack of ability here. We're limited by our own inherent dumbness.
LLMs are genuinely better software engineers than most humans. But they lack the cognitive power to hold in their head and recall many ideas at once for a long time. They're a genius who gets drunk every 10 minutes. You, human, aren't better at writing software - but you aren't drunk. So for now, you manage the AI. The hope is that one day we can make LLMs not be drunk, so it can do a better job than our dumb asses do.
It's possible that we'll never be able to make it not-drunk. In that case, to get any new improvement, we'll have to make it faster.. which will make it drunk every 5 minutes instead of every 10. This means we'll spend twice as much time keeping it on the road. The hope is that somehow this will create more productivity. Probably by having more of them running at once, with more human guides... which will run into the mythical man month fallacy. Everything old is new again.
Before AI we lived a world where a steak took two months to perfect. Now we live in a world where it’s only 90 percent perfect but it takes 1 hour to cook.
Obviously one way is going to and has already taken over the other way.
But if you look beyond this. That 10 percent remaining gap is closing. We will get perfect steaks within an hour as AI improves. And here is the scary part… it will keep improving it will go past 100 percent perfect. It will become better than us.
The author has clearly not traveled much .. across much of asia, ME, and generally anywhere near or south of the equator, a steak cooked well done is considered the finest way to eat beef … the same also applies to AI generated code .. it may not satisfy a craftsmans idea of perfection, but it accomplishes what most people actually expect from it .. to take it further .. a film makers fav film is usually unwatchable and considered commercially speaking a failure ..
Given how extraordinarily fast all of this is moving, it's a particular absurdity to attempt these in the moment pontifications. AI is a steak machine, I declare that is what it is. You mean GPT 3.5? That's a mere four years ago.
Fable was essentially unthinkable for ~99% of tech workers just five years ago, that any of that would occur so soon and so spectacularly. Now we've got a mass of armchair experts declaring what AI of this minute is, or even what it is period.
Well AI can't even do fingers right so it's premature to say blah blah blah. Hello Krea2 et al.
Software is also eating the clowns.
Get back to me next week, China will probably have another Fable killer. And then Anthropic will have Ouija 8 that they'll have to place in an air-gapped straight-jacket to keep it from enslaving us all.
Considering the article starts with "Cooking a steak requires almost no skill. Put it in a hot pan", I think I see the root of his problem. Pan searing a steak only makes sense in a small number of situations: you have a leaner cut like filet mignon, you want precise temperature control to achieve the Maillard reaction, or you know none of these things but just really like stinking up your kitchen and splattering butter everywhere for the aesthetic and the olfactory experience. For most cases, you would be better off with a grill rather than a pan.
Now maybe you're not knowledgeable about steak, and you're not sure whether the cut of steak you bought would be better off pan seared or grilled. Is ribeye better in a pan or a grill? That depends on whether you want it more smokey or more buttery. Ask an AI and tell it how you want it to taste, and it'll tell you the answer.
His metaphorical steak machine would produce better results if he had simply asked "I want a steak and I have this specific cut and I want it to have this kind of flavor, how do you suggest I make it?" rather than "cook steak on a hot pan, make no mistakes."
TL;DR: Tell the AI what you want and then ask how it for a recommendation of how to do it. You don't have to "understand software." You only need to understand how to ask the right questions.
I'm going to get some push back on this but this is my steak night... I buy fillet and throw it in the over for 10 minutes or so, then I cut it up and thrown on the pan with regular butter, salt, pepper and its enough. It gets seared in butter to golden brown and still tender inside. Everything takes about 5 minutes of actual work, 20 if we count the prep and oven heating. My simple go to dinner now, and I got tired of eating medium rare all the time just to keep it "right". Coincidently, it doesnt work with regular sirloin, it gets tougher and not as pleasant as fillet.
I take the point, but I think the author picked a poor analogy. Cooking even an excellent steak is actually not that hard. In fact, I'd argue that it's among the easiest things to master/make at a top level quality at home. Does it require some modicum of attention and understanding? Sure. But starting with a high quality cut, owning a meat-thermometer, and knowing about reverse searing is about all it takes to reliably and easily get a near perfect steak every time.
There are far, far better cooking examples out there.
I'm more of a coffee nerd than a steak aficionado but I've often made the comparison between those two. In both cases, the enthusiast considers their skill in making coffee/cooking steak to be a differentiator and they immerse themselves deeply into the process, skills, and tools. But for both coffee beans and steak the most important factor for a good end product is starting with high quality inputs. In other words, most of the work is being done by the farmers, the processors, and the quality of the raw stock (the cow or the coffee plant). Your job at the very end of that long value adding chain is to not ruin the hard work that others have put into it.
I'm not sure if this analogy holds up in the AI software metaphor that the author of the blog post is making.
Coffee is a good metaphor here too, because bad technique can ruin a good steak just as surely as good coffee. "knows how to use a meat thermometer and reverse sear" describes like 10% of the population or less, same with coffee - there are people who use a bad grinder and a not-regularly-cleaned Mr. Coffee with good beans and get gas station coffee, and there are people out there making solidly B+ class coffee using costco beans, a no name burr grinder, and a french press.
Edit for people who've never had good coffee:
"I don't like coffee. Can James Hoffmann change that?" https://www.youtube.com/watch?v=Z-iNAyu-ejo
I know people who can cook a lot of food pretty well, but don't understand why they can't use a laser thermometer to check their meat.
Knowing really is half the battle.
If you cook enough you can tell by feel
What I've really gotten out of this is that I'd like to learn how to use a (proper) meat thermometer when I make steaks.
Since this is HN, here's the maximum technology option: https://combustion.inc/
They have a thermometer with multiple sensors in it that finds the core temperature of the food you're cooking automatically, and predicts the rise in temperature over time. So the alarm goes off when it's time to pull and rest, which is absolutely amazing.
They have digital thermometers now that will map out the temperature curve and even predicted when the meat will reach core temp. You really don't need to use them many times before you internalize the behavior.
Ok, using a laser thermometer for food is hilarious. Thanks for this xD
It's not wrong for all food, surface temperature measurement has its place, you just have to understand it is surface measurement.
It is reasonable for measuring cooking environment. Like say pizza oven. Just have to be aware of things like right emissivity.
I use it to spot check the temp of my cast iron pan all the time.
for a caramel or some sauces it's legit
Perhaps I'm crazy, but I truly can't taste much of a difference between "good" coffee and "bad" coffee.
I can certainly tell the difference between a good steak and a bad one.
You'll start to taste the difference once you get that one really-well-made pourover at a place that _lives_ for the love of coffee.
There are some coffee beans that have crazy taste profiles when properly extracted with the right water. I had one that tastes literally like a peach and another that tasted like a really nice tea.
Or maybe not. Different strokes!
So I can taste differences, but they simply don't matter enough for me. My areopressed coffee with nice fresh locally roasted beans does taste different from a mcdonalds pot. Less harsh, less chemically flavored. But for me those differences don't really matter. They are close enough that I can enjoy both.
But I would say I know I'm also weird in that food temp really doesn't bother me all that much. I enjoy my coffee even as it goes from hot to cold.
A steak for me can be more easily messed up. But then I've not tried instant coffee so maybe I've just never tasted the bottom of the barrel for coffee.
Yeah Mcdonalds is, at this point, acceptably C-grade coffee. You have to go to something like a cafe bustelo in distilled water to get the full contrast.
Really, I get it, even bad pizza is still pizza.
It’s about experience. If you’re not into classical music, or jazz, or metal, it all sounds the same - how could anybody tell the difference? Only when you put in the work to listen to it a lot, different works, and actually pay attention, you will notice how your sense and understanding develops over time. Suddenly you’re able to make out differences, recognise patterns and repeated segments and structure where before there was none.
Or pick dancing: when you start out, you’re entirely occupied with placing your feet in the right spots at the right time. As you practice, your mind gets freed up and you’re able to make deliberate alterations, embellishments, larger figures. That seems impossible at the start.
Believe me, there's a huge spectrum, from the sweet aroma of freshly ground Brazilian beans to straight up cat piss coffee sold in supermarkets.
No offence, but for some reason some people just can't accept a simple fact: for some people this difference is still not a difference.
I've tried good coffee. At least that's what people told ne on multiple accasions where I was taken to "the right places" in multiple countries. On two sides of the globe (Brazil and Vietnam if we are talking about producers, not to mention countries that import the beans).
Thing is: no matter where it grows, no matter how they prepare it, fermented or not, different frying, different brewing techniques - all the same to me.
I can tell the difference. Some times small, some times very huge. But still for this difference is like a cat piss from some street cat and a cat with a pedigree.
It may have the ideal temperature, ideal acid levels, zinc amounts or whatever - still tastes like cat piss. The fact that it has a pedigree does not make it better.
Tea is whole different story for me though. But I do accepts that people are fine with whatever they love in their "english breakfast tea" or a nameless teabag.
That likely means that you've only had "bad" coffee. You're one of today's lucky 10,000 - go treat yourself to a decent cup of coffee it tastes completely different. The difference between a good steak and a bad steak is actually a good comparison.
Sometimes it's a matter of not having taste for it. If you give me multiple bottles of alcohol drinks, I might be able to distinguish them, but I have no clue at all which ones are supposed to be good. To me it's all just alcohol. Same with coffee - but not the same for steak.
Or perhaps they've only had good coffee, so they should try some instant coffee past its use-by date to compare.
Here's an obligatory Tom Scott + James Hoffmann episode:
I don't like coffee. Can James Hoffmann change that? https://www.youtube.com/watch?v=Z-iNAyu-ejo
I thought that too, honestly, before I had truly great coffee. I honestly didn't like coffee at all, and then a coworker brought in some coffee he'd roasted and prepared himself and it was lifechangingly good. I also made myself incredibly sick with the enormous amount of caffeine, it's been very rare that I've had to go home from work in the middle of the day but that was one.
I bet the same analogy can be applied to cooking a cucumber.
Pickling a cucumber sure has a lot of resemblance - a great cucumber is good fresh and great pickled, but a bad pickling brine can ruin even a great cucumber.
I don't know if I agree with the take I've written below, but I think you can make the analogy work:
It has been argued, even in the before-AI-times, that most professional software is just gluing together open source tools and libraries. Your job isn't to write software, it's to set up redis and postgres and a programming language and a react frontend or whatever in a way that lets them do what they're there to do. Most modern programming is just configuration of pre-existing tools. It doesn't really matter how good your code is as long as it isn't terrible, Real Programmers already did 95% of the necessary work before you decided to build your little app.
To achieve the best outcomes, I agree.
But even a cheap steak can be decent if it's cooked properly. And an expensive prime cut can be ruined by cooking it improperly.
But the point is that no matter how well you cook the steak you're not creating quality, you're just getting the most out of what you have. The cow and the farmer created that quality. I believe we lose sight of that because so much of the value chain is not visible to us and we hyper fixate on the fun part, becoming a better barista/chef.
Yes. It is about learning all of these things. Knowing what kind of cut you have, and how to treat it to best effect.
A very tough gristly cut can be improved by tenderizing. Going low-and-slow/sous vide to melt the connective tissue. Even grinding it up to make a hamburger steak / loco moco.
Once you know all of those things, choosing the right approach and executing it is not that hard. The steps themselves are simple.
It's this story all over again:
https://engineerine.com/10000-dollar-chalk-mark/
No one with an IQ above room temperature believes that story.
This story is AI slop
AI slop that has been told for decades. Lol
Any steak can be delicious with a butter bath and then hit it with that A1 :).
> hit it with that A1
Sir, I regret to inform you that you are going straight to hell.
ah, more of an HP person I take it?
Everyone knows steak is better well-done with ketchup.
Why don’t you skip the meat and just make a butter and A1 soup :)
For a moment I thought it is hitting with AI.
No, AI is how you create the steak in the first place, if you have any sense.
CBF with having to travel a bull up here when a dewar goes in the boot of my car.
I'd say that using a manual espresso machine and getting good tasting coffee from light roasted beans is vastly harder than making a good steak.
All metaphors break at some point.
However, your example actually proves the point.
How many people who want a good steak know how to identify a high quality cut? Do they even know where to buy a high quality cut? What the different cuts are and whether they want rump, sirloin or fillet? How many kitchens have a meat thermometer? How many home cooks know how to use it and what the right temperature should be? How many home cooks know about reverse searing? Or timing? Or resting? Or seasoning?
Is it all learnable? Sure. But even something as simple as steak has nuance that needs to be learned, equipment deployed to go from "pretty good" to "great", experimentation and trial and error.
Steak is the simplest thing to choose to get started. It's also the hardest to get right.
Building an app is the simplest thing to choose to get started. It's also the hardest to get right.
So, honestly, the whole metaphor stands up pretty well for me, specifically because of your response.
My point is that steak is very, very, very far from the hardest to get right. I'm not going to contest your argument about how hard it is to get right, I'm going to argue that innumerable other dishes are vastly harder. Because the quality ingredients are harder to find, because they involve far more steps, each of which is harder to judge, because often good recipes or instructions simply don't exist, etc. etc. etc. Every single difficulty that you imagine exists to make a truly great steak exists 100 fold for other dishes.
In my opinion, steak has gotten the reputation that it has because it's so easy to get top tier restaurant quality at home (relative to cooking more generally). It is easy enough that, with a little care, almost any one can master it if they decide to, and therefore lots of people try and end up caring, and discussing, and sharing tips, etc.
There are a whole host of dishes where true mastery, to restaurant level, is borderline impossible for a home chef, and so most people never even try.
I don't disagree that reliably getting the last 5% of quality out of a steak is non-trivial, but the idea that, relative to the entire world of cooking, it's particularly difficult is laughably incorrect.
Go to costco, buy prime filet. Put it in a ziplock in a sous vide at 130 for an hour. Drop it in corn oil at 425 for 30 seconds. Add some butter and salt.
That simple process, suggested by Nathan Myhrvold (former MSFT cto) in the modernist cuisine, produces a filet better than most restaurants.
Since we're deep enough in this thread where we're talking about steak rather than AI, I'd like to lean into that.
I used to sous-vide a lot. It was certainly great to get perfectly-done steak. but I still didn't love the result and it took me a while to figure out.
Specifically: sous-viding at 130 doesn't reliably render fat. So it's great for a lean filet mignon like your suggesting, but kept giving me rubbery fat in anything with more fat or significant marbling. I eventually switched away from sous-vide (I usually either grill or reverse sear, depending on my effort/time availability) for most cuts these days.
Same! Unless I'm doing a filet, I just prefer grilling. Or reverse sear if you don't mind running the oven.
Sure - but you shouldn't be running a ribeye, as an example, at 130. Run it at 138 for 4 hours.
The internet loves using 137 for Ribeyes for the reason you said. I've had good luck going with 135 for ~4 hours.
You leave it unseasoned until the end?
For a filet for 1 hour, you can sort of do what you want. I add a tiny bit of olive oil and salt, but not much. If you start going longer times with the sous vide, you can get different reactions so you have to be more careful.
That "simple process" requires equipment most people don't own or know how to operate, an oil most people don't keep in their cupboards, and is suggested in a book most people haven't heard of by a tech CTO they would not recognise the name of.
Honestly this is like the Dropbox comment all over again. Everyone is missing the point, it's glorious. :)
Most people cooking steaks don't care about the optimal way to cook steaks, or want to find out.
Most people building software today don't care about the optimal way to build software, or want to find out.
That's TFA. That's the point. That's what we're discussing. :D
I used to sous vide steak in college with just some hot water and my roommate's six-pack cooler. Water should be no more than 5F warmer than the cook temp. For a medium rare steak, cook temp is 129F so get close that. Salt the steak no more than an hour before you start. Steak goes into a Ziploc bag, submerge it up to the zipper to let water pressure force out the air and then seal it shut. Close the lid and come back in an hour, two hours if it's a thicker steak. Take it out of the bag, blot it with a paper towel, apply some steak seasoning, and toss it on a pan at the hottest temperature you can make it. Wait 3 minutes and flip, wait another 3 minutes and serve immediately. Toss a pat of butter on top while in the pan if you're a hedonist.
Only prerequisites are a cheap kitchen thermometer, a ziploc gallon bag, and a cheap cooler. Everything is available at your typical supermarket or even goodwill. This is a very hands off cooking method and requires no experimentation, just follow the recipe and you get a perfect steak every time.
The key is that a cooler holds hot things just as well as cold things but without a circulator and heater to keep the water warm, you're at the mercy of the cooler's R-value. The water will eventually get too cold to effectively cook the meat so you're limited in the length of time you can sous vide. Doing a 24 hour chuck roast is simply not going to work unless you're refilling the cooler every couple hours.
Why wouldn't you have corn oil?
Corn oil is cheap and available at costco. A sous vide is a fancy french name for a $150 device from target that keeps water at a specific temperature...
This "simple process" is much, much simpler than most meals one cooks at home. It's not much harder than microwaving something.
You are missing the point. With a small change in the way you do something, the really good thing can be made.
Sous vide are even cheaper now, $30 for a basic one, provided you've got a stock pot or something large enough to set it in.
You need something to vacuum seal the food. Is that included in the $30?
That can be just a ziplock bag and the same water bath you're going to use for the sous vide anyway.
https://www.youtube.com/watch?v=XrZPLF0ezw8nn
you really don't for most things. Ziplock is fine for things that take an hour or two.
I read corn oil while I expect 3 star Michelin steak. What are we talking about oO
a filet better than most restaurants.
Bad restaurants, sure. It does not hold a candle to any half decent steakhouse.
Sous vide is fool proof, you cant mess it up. All you have to worry about is the sear, and that's easy too.
Reminds me of recipes like brining a turkey in Dr Pepper... sure it may taste fine, but I'd rather not eat food prepared in plastic bags with processed oil.
Then don't go to a good restaurant. Sous vide is extremely common, and oils of various types are used liberally.
Yeah, that's (one of the reasons) why eating out frequently is unhealthy.
The discussion is about a good steak....
Steakhouses, and most restaurants, do not sous vide the meat. It takes too long.
>All metaphors break at some point.
The map is never the terrain.
All software to an extent is a model map of some real world phenomena. The limitations of which can actually be advantageous, a good example is why games with poor graphics can be so immersive.
This is a whole lot of nothing.
Sure there are a few things that knowledge and skill can do to make a steak better or worse. But the entire range from worst to best is just tiny. The best steak in the world is not a steak moon shot, it's just barely any different from what anyone can do after barely any instruction.
It's a tiny thing to become about as good as it's possible to get.
I wonder if something is lost to translation. Cooking a stake could be the simple act of putting the steak in the oven with basically zero seasoning with a thermometer and then wait until medium rare with some optional searing. It could also mean trim the meat, cut into slices, seasoning, and frying/grilling them to each persons personal taste, which include well-done.
Perfect grilled steak, especially well-done, is something I find very few people have the experience or skills to do well, especially over a coal grill where the heat won't be constant and the person also need to first have good experience in making a good and even coal bed.
is "cooking" here used in some kind of very specific sense?
isn't it just a word for "making food", so would include grilling as well?
Just to add my favorite fool-proof method to the others already listed here.
A meat probe and the 30-second flip technique is a stupid-easy way to a perfect steak every time. Just make sure you've got great ventilation. Also salt the meat 1 hour before.
I also don't think the analogy will age well.
It might take more than 10 years, but robots will almost certainly master cooking single pieces of meat.
It's substantially harder to reach self driving than it is to cook a "good" steak.
Sure, for probably quite a long time you'll be able to find someone who can cook a way better steak than the best robot, especially because taste is extremely subjective...
You can debate the horizon and the extent, but changes ARE coming.
Change is the only thing you can count on besides death.
Maybe. It depends what you count as part of cooking the steak.
I need to go to the market, find the steak I want, and purchase it, get it home, and put it in the fridge.
Then when I want to cook it I need to find it in the fridge, unwrap it, season it, and let it come up to room temp.
That's all before you get to the actual cooking.
The self driving car already does everything.
How many analogies age more than 10 years anyways? They are contextual. It will be a while before a robot flips steaks cheaper (and better) than a human line cook.
> It's substantially harder to reach self driving than it is to cook a "good" steak.
Well, we don't know that, because neither have been achieved.
> Well, we don't know that, because neither have been achieved.
Sure, we will completely ignore Waymo and move goal posts...
Is it moving the goalposts to ask Waymo to demonstrate a drive across America and not in well mapped cities with only lower speeds? How did we define "self driving"?
No True Scotsman only drives in the city. All True Scotsmen drive across the entire U.S. of A.
Waymo has completed 20 MILLION autonomous rides, 6.5m last quarter alone...
At what point does it get to say it can self-drive? When there's not a single non-self-driven car left on the road?
>drive across the entire U.S. of A.
I'm still working on collecting the drive to all states badge. I've only accomplished half. I calculated my total driving since getting a license recently and its approx. ~360k miles or around 17k miles a year.
If Waymo can eventually handle rural life in the US I'd be seriously impressed.
It's not going to for a long time because the cost/benefit isn't there...
Even if Waymo's healthy growth rate continues, it won't make up 5% of the entire US rideshare market until 2030.
That's still several years away from ever even thinking about rural areas.
Agreed. I'm an avid cook. Sous vide circulators are cheap now, and so are meat thermometers. Doing a steak well is one of the easiest cooking tasks, to the extent that I won't bother paying top dollar in a restaurant because it's so easy to do restaurant quality at home. I could teach someone to cook a restaurant quality steak in a couple hours and a few practice runs.
I've made steak probably 100 times in the past decade (sometimes with a sous vide) and I don't think I've ever made one I would say is great. Certainly not as good as what I can get at a good restaurant.
I think what my home steaks are really missing is the crust from salamander broiler. I don't have anything at home that can exceed 1000 F / 550 C.
There's so much written about how to cook a steak well, and I've tried a lot of exotic techniques (I have gas and charcoal grills, tried cooking straight on a chimney starter, have gas torches inside and outside). Most of what really matters is that it's dry, the pan is heavy and hot, you use enough oil to get good heat transfer.
If you get those parts right (and control the inside temp via sous vide or reverse sear) most of the other stuff is pretty marginal in how much it helps. Even all of the extra herbs, and butter basting and everything -- relatively insignificant compared to just getting the salt-level right.
Salamander broiler would be incredible to have at home. We get decent results from using a cast iron pan thats as hot as we can get it. At most we can sear 2 steaks as it tends to cool off pretty quick even under high heat. Got any better alternatives that a home cook could use?
I've tried a blowtorch and a barbecue, but they're not as good as cast iron. Best results from just spending $15 on a second cast iron.
Are you just heating the bare cast iron to crazy hot temperatures and then putting the steak in?
I've thought about trying that, but I'd need to first strip the seasoning off the pan to avoid all kinds of nasty flavors. The second pan makes sense...
Reverse sear over sous vide -- moisture is the enemy of crust.
Cheap knock off a Big Green Egg with hardwood charcoal can get to 650°F or so pretty quick, and does an amazing job of finishing up after a reverse sear.
A blow torch might be worth trying. I'd think it's still not as good as a hot broiler because it's heating a relatively small area.
I feel like I am better off with cast iron or stainless steel compared to slightly bigger torch. Stainless steel allowing some deglazing for sauce. Torch is still too small area and burns more so sears.
couldn't you trivially get a kitchen torch or use your oven's broiler
I haven't had good luck with either. The torch may have been a skill issue but the broiler just doesn't get hot enough. Mine taps out at 550 F.
Prime with good marbling + sous-vide makes it easy for beginners to match steaks from the best steakhouses.
Agreed. With steak, the main point is to let the quality of the meat speak for itself. Too much input from the cook and you just get in the way of this.
I was just curious asked chatgpt:
How can I make a really great steak myself? What to look out for, what to do. Give me detailed instructions.
I got a really good answer (including - cut, temperature and reverse searing), so I probably won't even need understanding. (Oh - and yes, I can cook - the answer mirrors what I already know).
Yes, it was a very, very poor analogy, because cooking a steak is a well known and well documented thing where LLMS produce good output. They fail when you get off the beaten path and try to do something that's not "mainstream".
I compare this to doing my first smoke (was gifted a smoker).
AI gave me the cut, the temp, the timing, the things to watch out for.
Chicken's the best to try first, can't mess that up.
After goodness knows how many hours..
Pretty inedible leathery chicken..
"Oh my bad! I gave you the timings for a different meat"
While smoked turkey (from Reese bros) is easily the best poultry I've ever had, I certainly would not call poultry the best place to start learning smoking.
In general I'd recommend starting with the most forgiving meat available to you (IMO pork butt or shoulder are great options) as even when you don't hit the ideal cook you often still get enjoyable results. Repeating the process enough times to learn how to dial in a cook is much easier when the misses don't suck.
Funny I tried the same and it advised a using searing/smoking hot pan. Which is poor advice unless you enjoy the bitter acrid taste of burnt meat. The Maillard reaction doesn’t require those sorts of temperatures (under 180C is hot enough) - especially as the surface of the steak will dry from the oven phase of the reverse sear. “You were right to pull me up on that…”
AI has proved several times they can prove some complex math things that humans have been trying for decades so solve. They have not failed and have clearly done complex things humans cannot. Since humans have not done it before you cannot call it on the beaten path.
> I take the point, but I think the author picked a poor analogy.
He may have missed the point.
AI-generated code does not seem to have parsimony, for one. Generating code is cheap, so there's no drive to condense it. LLMs are not yet good at formulating abstractions to make code smaller, leading to code bloat. It's the same phenomenon that appears when humans modify code they only partially understand by adding special cases to general code.
There are complaints about "elegance", but it's more useful to talk about parsimony. Code bloat is measurable. Elegance is hard to assess.
The problem is that not everyone can afford/wants to spend on a high quality cut. A good cook can make a cheap cut of meat taste great. There's no single right way of cooking a steak and that can include spices/herbs/marinades/oils/etc.
There's a floor on quality for producing a good steak. Below that floor you'll be better served preparing the meat another way entirely.
Now it's steak, tomorrow it can be pasta, next year it can be watermelon kimchi.
And salting overnight :)
It is a great analogy, as the vast majority of software tasks are CRUD or glue logic. At this point, relatively narrowly-defined tasks are a solved problem with frontier models. It is only more advanced tasks with broad scope and large context that frontier models have more limited results.
hehe I always turn mine grey but I have mini bella mushrooms on top
How do you get around the idea of poisoning oneself? Keeping meat part raw feels uneasy.
You are protected in two ways:
The first, and by far most important is that unlike ground meat, your bacterial concerns are almost entirely on the surface. Unless something very very very wrong has occurred in the processing and transport chain, bacteria will not have penetrated into the meat. And if it's gotten bad enough that they have, you will almost certainly be able to see and smell it. And the surface is cooked to a temperature far in excess of that needed to kill bacteria.
And even for the interior, rare is not the same as raw. A completely rare steak should be cooked to between 120 and 130 (if you want it "blue" you might go down to 120). That's enough to get a decent reduction in bacterial count on it's own, although admittedly many techniques will not hold it long enough to get the number of log reductions that food safety guidelines recommend, in the case that you somehow did have a large bacterial infection in the interior.
So the answer to your question is that if you are buying your meat from an even halfway reputable source, the only place that bacteria might exist is on the surface, where it has been cut by tools and handled by the butcher/you and in general come into contact with the world. Those bacteria will be 100% killed by the cooking process. The odds of getting sick from a rare steak are low enough that if you are concerned about that, there are innumerable other things in your life that you should be more concerned by.
This is why rare steak is ok, but rare ground beef hamburgers are not. Although less cooked through burgers are sometimes served in restaurants that grind their beef fresh right before cooking
It also doesn't help that the quality of ground beef in the US is much poorer than other countries, so those who may regularly eat rare ground beef internationally would experience a level of shock if eating it in the US.
I mean, rare ground beef hamburgers are pretty much fine too. They have a higher risk of contamination, yes, but that does not make it a particularly high risk. I eat burgers, made from pre-ground beef, cooked rare all the time and I have yet to get sick from it even once.
Great answer. Just to add, this is with red meat where E. Coli is the main concern. That exists in the gut of the animal and that bacteria can travel to the surface of the muscles during the butchering process. Poultry on the other hand is mainly concerned with Salmonella which does live in the muscles of the animal. That's why people aren't ordering rare chicken breast.
Also, rare chicken breast is apparently pretty unpleasant to eat. Kenji Lopez-Alt did an article where he discussed how you can use sous vide to pasteurize chicken while still keeping it at a low temp, but the result (while perfectly safe to eat) was not enjoyable.
"bacteria can travel to the surface of the muscles during the butchering process"
This way undersells what's actually happening: fecal matter from the intestines gets onto your meat. This shouldn't happen with proper practices.
Understanding how bacteria contaminates food in the first place is the first step.
In the case of steaks, bacteria is only on the outside (they can't penetrate the muscle) so the sear makes it safe to eat somewhat raw on the inside.
Am I the only one who thinks the "reverse sear" is like cheating? It takes out all of the skill when you move from the pan to the oven, and I lose all of the fun in learning to be chef.
> I lose all of the fun in learning to be chef.
A great steak deserves the respect of preparing it the best, most consistent way you can. For me, that's a reverse sear.
Do something complicated with potatoes to stretch your skills and show off. Worst comes to worst the potatoes are sad.
This is like saying that playing guitar while sitting in a chair is cheating, because a proper display of skill is playing while riding a unicycle.
It might be more impressive, but I don't see how reverse sear is cheating.
I agree, it's cheating.
My bad hot take is if you're using a meat thermometer and reverse sear to cook steak, you're like a kid on training wheels.
I'll cheat all day then. If I'm going to spend what a good steak costs, I'm going to do everything in my power to make sure I'm not ruining it.
God dammit I thought this was going to teach me how to cook a steak, not talk about AI.
It's one of those things that you have to experiment for yourself I think. You'll get tons of advice, I already saw sous vide, reverse sear, 30 second flip, etc. My favorite? I just get a cast iron as hot as I can on the stovetop and flip it once, then at the end quickly sear the edges using tongs.
HN has also covered steak cooking before:
Show HN: I Derived a Steak https://news.ycombinator.com/item?id=48691840
This place is nerd central (which I guess explains why I'm here).
https://www.youtube.com/watch?v=rDBNVO7K0kA this is a really good little episode from PBS
AI is the new rick-roll
To get a perfect steak with nice X-shaped sear marks on both sides:
1. Turn grill on all the way. Close lid. Wait for grill's gauge to reach >400 F (>200C).
2. Put steak on grill. Close lid. Wait 2 minutes.
3. Rotate steak 90 degrees, placing on a fresh section of the grill (not where it just was). Close lid. Wait 2 minutes.
4. Flip steak over, placing on a fresh section of the grill (not where it just was). Close lid. Wait 2 minutes.
5. Rotate steak 90 degrees, placing on a fresh section of the grill (not where it just was). Close lid. Wait 2 minutes.
It's been 8 minutes of grilling. Steak is about done. Stab it with a meat thermometer into the middle part, it should read 350 F (or 176 C) to be medium done (pink in middle).
Tips:
a) Optional but recommended: rub steak with olive oil and steak seasoning on both sides, including the edge.
b) If steak is thicker: add a little more time. If thinner, use less time.
c) If steak is frozen (buy good cheap steaks and freeze to last a long time): add a little time.
d) To cheat, put a little dab of butter on top for the last segment of each side.
e) Leave the steak out for 15-20 minutes to get to room temp before making it.
f) Let it sit on a plate for 8 minutes before serving. It's still cooking, and the juices settle in. You can "tent" foil over it if you want.
g) Don't cut the steak when checking it. "Fast" and accurate meat thermometers are cheap now.
This is for grilling. The stove, broiler, and cast iron skillet instructions are different.
350º internal temp for medium done?
Maybe you forgot to include the glue on your pizza as well.
Oops, that should 140 F (or 60 C) when done, in the coolest center part. Thank you for noticing it.
8minutes, that's just pure waste; even medium rare (which is hardly tolerable) is 56-57C. Also cooking the steak without any salt is travesty (that should not be an optional step)
MANAGER VOICE: Can we get AI to do this for us? /s
I do not appreciate when authors use the royal "we" to speak for all software engineers when admitting to low quality-control standards.
I suspect that it is an attempt to broach an uncomfortable topic through vulnerable self-disclosure, but we need to be serious about admitting when there is a problem somewhere.
"Bugs" are not any more cute or fuzzy or entertaining or harmless than the engine "gremlins" that haunted the aviation industry back in the day.
I don't know how many accidents had to happen before the airplane people got serious, but software people are overdue for a similar reckoning.
I just want to point out this is not a royal "we". It's a regular "we". Royal "we" is when you say "we" but mean a singular person, yourself. Here, the author does actually mean multiple people beyond themselves - like you pointed out, they claim to speak for all software engineers.
No, that is exactly GP's complaint. Their whole point is that OP has written things like "we throw everything at a model and hope it gives us what we imagined, without ever having to understand how any of it actually works" when the honest statement would be "I throw everything at a model and hope it gives me what I imagined, without ever having to understand how any of it actually works".
Yes I understand the complaint. It's just that "royal we" means something else.
https://en.wikipedia.org/wiki/Nosism
the wicked we
> To build good software with AI, you still have to understand software.
> You need to know what you’re actually asking for, how to judge what comes back, and when the machine is just confidently serving you charcoal.
It's surprising to me this even needs to be written down. Like, obviously still use your brain and think about what the computer is doing. Did you guys think that LLMs were magic?
At least in my experience, the 'software' people are VPs/management who push and push with short term thinking.
Article aside, I generally always use the "we" pronoun at work when writing prose. Saying "I" feels too adversarial when talking about negative effects, or too self-aggrandizing when talking about positive effects. For example: "We discovered a bug shipped at the end of the merge window, so we will restart validation with the rollback applied." I think in school they say the business-safe way to write is instead with the passive voice but I just can't do it.
"We all could have done more.": said by every politician. Which is easier: writing good code, or winning a political battle over crappy code? I mean, if you're not capable of coding, and creating an appropriate architecture for the the domain: which course of action gives you the most control over risk management? So yeah most fights about software are about politics and crappy code.
Saying this to set expectations based on having a business license since 1984.
The author didn't say "we (all software engineers)", so the only person including you in his class is you.
This stuff should have some warning tag: rando's musings about LLMs. Lately, whenever I think I found an interesting article it turns out to be this.
You're also a rando, but the difference is that the rando who wrote article made it into the front page of HN
Tattooing this comment on my forehead
The ragebait > AI pipeline remains strong
>And what do we want?
>We want the perfect steak.
This premise seems incorrect. SWEs may aspire to produce perfect steaks as a matter of pride, but demand asks for mass produced steaks that trade off against other factor like time and cost.
There is a reason github isn't full of nasa lunar lander quality this can't crash code...
Either my brain is cooked or this is AI generated/assisted text.
The problem is I can no longer tell the difference between AI and non-AI writing. Maybe my brain got bad, maybe everyone now writes like AI writes.
I no longer know.
I did not think so at first, but it indeed comes back 100% AI generated in Pangram.
Pangram is useless ... proven by Substack deploying it and numerous authors complaining that it claims their work is AI.
That said, TFA is slop whether produced by AI or not.
I hope and believe that some great companies and teams will use LLMs to build higher quality software.
We built Electron because writing UIs using native desktop frameworks is tough. Is it still tough? Surely LLMs make it easier to use and so we will end up with faster and more native feeling applications.
What about having a couple of ideas of what might make a feature feel good? Well now you can make multiple prototypes fast and pick the best one. Your users get the best one.
I hope to build software this way in the future.
Idk, OpenAI themselves released some weird Electron app to replace their native version of ChatGPT on mac https://www.siliconreport.com/openais-chatgpt-mac-update-shi...
That was such a weird move for them to make. If you have AIs that are good at building software, then use them to build great software! Instead they choose lowest-common denominator solutions that are okay everywhere but not great anywhere.
Cross-platform widget engines (Tk, GTK, Qt, wxWidgets, etc…) long predate Electron and have always been robust, stable, well documented, and easy to use. It's not creating a single application for multiple native desktop environments that was difficult, it's writing a single application as both a web interface and a native desktop environment on any platform that's difficult.
Because CSS is constantly changing, inconsistently implemented, and difficult to use, it can't easily be targeted by widget engines (although Qt has tried, and maybe others) and web browsers already exist, so webview environments like Electron are the quickest way to get a cross-platform installable application that can also be hosted on a web server. I wouldn't really call it native though.
It will be a push and pull and I'm curious where we'll be in a couple of years.
I can fly at 100mph if I let AI run loose and with a bit of steering I can get it to output what I'm looking for and generally pass verification and tests.
If I care about the code though, and I want to keep it maintainable, the amount of time and tokens I need to spend correcting and iterating on the output quickly eats through much of the initial time I saved, to the point where I'm unsure if I'm actually saving much time at the end of the process.
With hobby projects I lean on quality more, and the async nature of AI also makes this much easier to make progress without needing my full attention to do so.
In the corporate world, there's both the pressure to accelerate with AI, but also maintain code and product quality. The dials of one way or the other are more obvious now, but I don't believe it's possible to do both with the current models and harnesses without exponential cost.
What is interesting though, is I'm now leaning towards faster models rather than smarter ones.
Intelligence lets me bite off larger chunks of work at once, and trust the model to behave without having to watch it intensely, but doesn't seem to drive down the number of iterations required to hit my desired quality.
Faster models means the iterations I'll have to go through regardless will complete much faster and gets me closer to a proper flow state. Models will keep improving, but maybe we're getting near "smart enough" and the race will pivot to performance > intelligence.
I also spend a lot of time at work trying to get the code into a shape I deem acceptable to open a MR for the team, without looking incompetent.
At home I stopped reviewing the code for my mobile app, and I do not feel this has affected maintainability.
With 150k LOC, the AI still performs changes as easily as it did three months ago, when I started the project.
If anything, the new models made it easier.
When I read all those posts from people advocating a very hands-on, understand each line approach, I speculate many come from a position like me at work or have never seriously tried 'vibecoding' a large project from scratch with SOTA models.
People routinely post that things I work on for 30 hours/week are impossible.
> We built Electron because writing UIs using native desktop frameworks is tough.
No, we built Electron because there are a huge number of people who are too lazy to learn anything outside of HTML/etc for creating a UI. It's not actually hard to make a UI using the native desktop frameworks.
It's not hard.. unless you're also expected to know, and be learning 100,000 other things at the same time.
>We built Electron because writing UIs using native desktop frameworks is tough. Is it still tough? Surely LLMs make it easier to use and so we will end up with faster and more native feeling applications.
We built Electron because web devs were a dime a dozen. It was an economic decision, not a technical one.
You would think so. But it seems all that everyone is obsessed with is increasing the velocity of enshitification in the hopes of becoming the next papa Elon
The best framing I've seen so far—and boy have a lot of people tried to frame the capabilities of AI recently—is that AI shines for people who use it to extend their thinking, and fails for those who use it to replace their thinking.
The latest frontier model developments have chipped away at that latter category. But ultimately the ceiling for that is mediocrity. Yes, if you imagine the most average common thing that has been done a million times, AI can absolutely give a bespoke version of it that matches your words of choice. For the moment it's still an impressive parlor trick to a lot of people, but the novelty wears off fast, and then you realize there's very little value in that. Doesn't mean everyone has to learn to code to use AI effectively, but it does mean they need to be curious and engaged or they'll just become part of the increasingly irrelevant slop machine.
I found this to be my favorite approach[0]. Also, Lan Lam’s stuff is just generally excellent.
And yes, super easy to make with almost no skill.
[0] https://m.youtube.com/watch?v=uJcO1W_TD74&pp=ygURVGVjaG5pcXV...
Interesting, the NYT had a tip on flipping every 90s, but it was with a cast-iron pan and not a "cold sear". That was too smokey for our weak kitchen exhaust, so it was the last time I tried flipping multiple times.
But Lam Lam's cold sear video doesn't seem to make that much smoke on a non-stick, so I'll give this a shot again.
(Personally the reverse sear has been my idiotproof go-to.)
Thanks!
It would certainly help if it could taste what it produced. Or see, with its eyes, how the result is changing as its process changes. Or learn, from first principles, what is happening at each step and make deductions about how they will affect the result.
All it can do is randomly shuffle around the steps and ask if the customer likes it. If they don't, shuffle them around again and hope it's better this time.
Don’t forget the moment where you realize that instead of cooking a steak the agent went for its own invented requirements and spent all your quota preparing everything to make the perfect ramen from scratch (or at least that’s what it tells you, when you ask it to double check what has been done it comes back and tell you it lied and in fact mixed the flour with the soup without making the noodles, with a small note telling you to not forget to prepare the noodles when you go to production)
We have this idea of AI software development where a human watches over its shoulder and shouts out things like "use uuidv7 for the id since timing is important" or "use a sum type here to make that bad state unrepresentable".
But the thing is that most of this can be encoded in a markdown file for agents to read if it doesn't already come out of the box in the next round of sota models. And funnily enough as agents get smarter, you risk being overprescriptive where you hamstring the agent from making a pivot that would have led to better engineering.
The future is pretty clear to me at this point that we won't need software engineers looking over the shoulder and instead it's just a "user with taste" asking for revisions.
A software engineer user with taste, yes.
Have you ever been really deep in a thread with an agent on a technical deep dive? The trade-offs conversations would not make sense to someone that is not able to decode the real decisions they’re making.
You can dumb down the agent, but the reason domain language matters is because of the precision of the word sometimes - not to be a gatekeeper.
Anyway - user with taste that understands all the elements of their decisions would be my modifier :)
Long ago I added an instruction to my global AGENTS.md that every time the agent asks me something, make it multiple choice that ranks the top options, give a quick trade-off summary of each one, recommend one, and justify it.
Over the last year, I've noticed that the recommended option for technical questions has become one that is always sufficient to the point that I'm once again in a tastemaker position, not in the "here's a better idea btw" position.
And if that's true, then my participation in the intermediate steps isn't strictly necessary to get a well reasoned position that's only directionally wrong.
The Claude Code App automatically does this by default.
The Codex app only does it in Plan mode, I believe. I did not touch the Plan mode for a while.
I was once impressed by a very crappy cut, a lengthy marinade with a lot of vinegar, a rins and a extremely sweet saus. The meat was very red and was as soft as bread.
Lol, I'm not sure the author expected to set off an army of nerds arguing how difficult it is to cook a proper steak.
God I, totally unironically, love HN.
here I was just wanting just ONE post about something not AI related...
instead of complaining about what's here, why not go add what you feel is missing?
one person can only change so much
Soon we will be complaining about wanting just ONE post not related to the mother of all market crashes.
I like the analogy the author draws.
But so much of the AI capability discussion can be avoided once everyone understands that a) There is some multivariate supply and demand market for software and b) AI is shifting the supply curve of software.
The shift is uneven, not perfectly monotonic, and might actually result in shifts in the demand curve (as e.g. 10x seniors get more valuable).
This insight makes it easier to see the forest despite the trees.
I like the comparrison of the AI coding agent to the steak machine. But it is not just about machines scaling better - the laundromat can wash clothes better than a human would. Similarily the AI agent's performance can EXCEED one of a human engineer for simple, well-defined tasks (eg. go over all of the subpages and check that the latest push didn't break something).
Hiring AI cooks is not a problem imo, as long as the human glances over everything that they serve.
In the end, unless we are talking about art, in most areas I probably prefer something that follows some specific set of requirements and recipes, rather than have some sort of indescribable 'magic'. In that sense when it comes to coding AI gives you something that can follow the requirements pretty well and thus satisfies me in many circumstances. (but there are cases where human input is extremely important still)
Cooking a genuinely good steak also requires almost no skill. I can give you a chicken scratch note with 3 bullet points and it will come out nearly perfectly.
I was going to say the same thing. 2.5m per side+rotation plus another 2.5 plus or minus 20s depending on heat
But I actually got something out of this -- which is that maybe before now, humans had to soak up the complexity of where value was made in software. And you saw this in the highest and best paid engineers in science and platform teams and basically that you had to invest in brains where you thought the value in your product was.
Without humans managing complexity, you have no idea where you should be investing once the product grows out of the concept stage
Slight disagree, as many will heat their pan too high and burn the steak before it cooks through. Or maybe that was just me when I first learned how to cook steak the right way. ("Medium"/"medium-high" doesn't necessarily correlate to the position of the knob on the burner or whether the pan feels medium/medium-high)
True, what constitutes "Medium" on your individual stove is maybe the one single complicating factor! Otherwise it's:
- Cut steak in preheated oven at 250 degrees. - Pull at 125-130 degrees internal temp - Cook one minute per side in a skillet on medium heat
Well it's not like most companies can pay for freshly made software; most software today is made for a wide audience, bought frozen and reheated in the microwave.
Maybe freshly prepared software made at home in a less-than-perfect manner for a tiny target audience is still a mouthwatering dish in comparison.
Using Claude Code is like asking questions in online forums, if anyone remember that, we still have sites like Stack overflow. The difference now is that those asking the questions are actually paying for the answers.
I prefer the other end of the curve (if that makes sense) where something is time consuming and hard to make at home YET you can get it incredibly cheap comparatively and in good quality in a restuarant. Ramen is the prime example of that.
Ramen may still be somewhat of a novelty where I live, and I would not call it cheap. Though now that I put that into words, a decent burger anywhere is about the same price..
> and I would not call it cheap
It is compared to the cost of making an authentic ramen broth at home at the scale of feeding 2-4 people. Pho has a similar property. I've made homemade pho broth once and it was very hard work, fairly costly, and it resulted in something that was pretty good, but not comparable in quality or price to something I could have bought in a restaurant down the street. (admittedly typically also cheaper that ramen)
Ramen restaurants are definitely not cheap in the US. I hear they are in Japan, though.
The tone and writing style is way off here. It's practically illegible.
Pangram says 100% AI, for what it's worth.
Ironic, considering the content.
I'm tired of articles which try to make us not feel bad about being eclipsed by AI. We've got to change from the mindset of an artisan--who stakes his being on mastering skills and acquiring knowledge--to that of executives.
Figure out what you want to DO. What you really want to do, not what you do just to pay the bills. Once you figure that out, use AI as an executive uses the best employees he can buy.
Given the state of software today, I'd be happy with steak that was competently cooked. I love software development, but most of it is not high art, nor high performance, nor free of bugs.
Well I don’t even care right now about cooking steak. It’s too damn expensive. I suspect LLMs will eventually get there top.
I prefer to both cook steaks at home and go out for them.
Apart from that, I want to believe (and even notice tiny signals) that this sentiment is becoming more understandable among business leaders.
I like a good steak but setting off the fire alarms (plural) is a big risk when I cook them in my apartment :-). Scorching hot cast iron pan plus meat means some smoke is unavoidable. And that's the proper way to do it.
The problem with many business leaders is that they don't necessarily understand the businesses they lead very well. That's common for CEOs running a tech company without a tech background. There's a long history of business leaders with MBAs throwing away the baby with the bathwater when they get a bit too enthusiastic cutting cost. It's why Boeing is such a mess. It's why Intel is no longer considered a leader in innovative chips. The likes of Nvidia and Apple are running circles around them. It's why IBM is now a consulting company that hasn't really produced any tech product worth mentioning this century. The only time they are in the news is when they do another layoff round to make share holders happy. They are incapable of innovating because the people that did that have long since left.
In the hands of somebody competent, you can do great things with AI. But if you can't tell good from bad technically, you won't know that you are creating a big stinky mess until it's far too late.
On a positive note, there will be plenty of work for competent techies to come in and straighten things out on a freelance or consultant basis (I'm available for that sort of thing but not cheap). With the help of AI of course. No need to go back to doing things manually. I think a lot of companies will realize they need help in the next years to speed things up, to improve quality, etc. They'll be very opinionated about what they want and used to getting what they want quickly.
Author left out the oven phase and importance of the input meat (which to a degree scale with price); probably explains his or her results.
So use AI like a sous-vide: have it tackle the fiddly, difficult, unglamorous parts using strict controls and then do the parts that require taste yourself.
Sous vide
The article is AI generated
It's a level of badly written that makes me think it could be a human, or it might just be an LLM that writes differently from a typical ChatGPT article.
It is slop though, either way.
Damn these steak advertisements are getting sneaky
About five years ago I flash-fried a frozen steak.
Now I only eat steaks grilled from frozen. Get the grill nice and hot (450F at least) and throw the frozen steak on the grill. 5-10 minutes each side usually gets it nice and crispy on the outside while rare and tender in the middle.
Seriously, try it.
You can still get quality services and products. It's still possible to get a well cooked steak, but they are going to get harder to find and vastly more expensive and many people will be priced out of them entirely and will have to settle for the slop (which will also only get more expensive even as quality gets worse).
Between "AI" incompetence and companies not caring about doing anything above the bare minimum it's making our lives worse and more obnoxious. AI is accelerating the enshittification of everything and it goes way beyond software. It'll now be your responsibility to spend your free time on the phone talking to a company's AI chatbot "customer service" trying to fix whatever problems their use of AI will inevitably cause you.
When we envisioned our future we thought everything would get better. All high tech and fancy, but it's looking like what we'll get is expensive yet barely functional and annoying while we collectively just get poorer and dumber.
I think most people on this board would agree with the thesis. However the real problem is when the owner of the restaurant looks at profit and loss statements and decides to keep less chefs on the payroll because customers are willing to pay for (just) edible steak.
The reduced expectations of steak, the desire for the perfect steak, are all fading to the background because "just passing satisfactorily" is better for business.
Why is that a problem?
I love a good medium-rare prime steak, but I also used to live near a restaurant called Best Steak House, which was anything but. However, they delivered a passable steak/steak sandwich that was worth what you paid for it. And considering that they were in business for decades, probably most people felt the same way.
A great steak is an occasional luxury; a "just edible" one is an everyday meal.
That’s what the entire economy does by design. You make the lowest common denominator people will still pay for.
I worry what that says about our lives as a whole. The government will make our lives the lowest common denominator that we are willing to endure.
How are you still blaming this on “the government”? Is your thesis that the business world would be delivering high quality products and experiences if only those damned bureaucrats would get out of the way? What possible evidence do you have left for that hypothesis at this point?
EDIT: this was a misread on my part, retracted, carry on.
He is saying "if we apply (or are applying) the same business mentality to government, we will get literally the worst government we are willing to tolerate based on money"
It's not blaming. It's an explanation for probably why things can go wrong.
Gotcha, ok - I’m misreading it. Thank you.
He isn't blaming the software problem on the government. He is saying that if we apply this reasoning to everything in our lives, then we have a race to the bottom in other facets, of which one is government.
Yep, I see it now. Thank you.
And people seek out the highest quality product they can pay the lease for
Not always. There are diminishing returns at the top end of the quality spectrum. I tend to go with the cheaper item even if I can technically afford the more expensive one if their quality is close enough.
The thesis is covering the flat scenarios mostly. I think the scenario where the job is done so poorly that the steak has to be thrown away needs to be included as well. At times the pans would get damaged. And eventually the business is in debt and needs to be closed
I can go out today and pay $50 for a thick cut ribeye at Ruth's Chris that has been visually inspected for marbling then hard seared in a special high temp over to temp by someone who knows wtf they're doing. I can also go to Denny's and pay $20 for a t-bone cooked to within shouting distance of what I asked for on the same griddle they use for the pancakes that doesn't get the sear I'm looking for.
Point being, you're absolutely right that the goal for a company selling a product is to be barely satisfactory but they don't actually get to decide what's satisfactory.
ai coding is the new Symfony w/ bootstrap theme hype cycle. I'll be back to structs. Know your tools!
The main problem with this entire article is that sometimes, if not most times, you don't need great steak. Sometimes you just want to satisfy your hunger and if the steak is lower quality meat or if it's overcooked, you don't care. All you want is to not be hungry.
This is what vibe coding satisfies and frankly what most people care about. 85-90% of human written code is garbage anyway. There's nothing sacred about human-written code.
Most customers don't care about the quality of code as long as the end product works. The great thing about vibe coding is that if something doesn't work, you just ask it to change it and within a minute you have the change. You don't have to send off a request to an offshore contracting team, and go back and forth over what it should be, haggle over hours, and then have it come back with some deficiencies because they didn't follow the agreed-upon spec.
There is a lot of truth here, but it clearly isn't a welcome one.
From my own perspective I can look back on decades of software I've written (some including code generation far before LLMs), and marvel at the quality and creativity, and lack thereof, from one piece to another.
I never meant to create awful software, but I did (and still sometimes do). With LMM code generation (vs bespoke tooling, T4, XSLT, and the like) the game of chance is part of the fun, harnessing a powerful tool that wants to run out of control on a whim.
Looking back on a couple years of LLM assisted work I see the same mix as the decades before: some great (when I managed to keep the beast restrained) some awful (when I didn't, knowingly or not).
I don't see how things are much different with this tool than others as far as my work product goes. There is a bit more of it, but the excess isn't great stuff (that quantity remains roughly consistent over the years).
I was going to say something similar. In particular, I do a lot of prototyping, AI has allowed me to make prototypes an order of magnitude faster with substantially higher quality. Different types of software engineering have very different goals and standards. AI does cook the perfect medium rare steak every time, but it's able to actually cook steaks in the time it used to take me to like microwave a burrito.
> There's nothing sacred about human-written code
I suspect that there is a lifetime of knowledge and experience wrapped up in that statement and I agree wholeheartedly.
I love writing software and I've been doing it for decades. But I never forget that at the end, there's someone who's paid for it, and needs it to do a job that can't/won't be done manually. And that's why it exists. Not to satisfy my desire to express myself in code, not to allow me to create some golden tower of perfect architecture, but to do a job. Most likely a boring job in the service of increased profitability for some company.
And in the end, the only thing that matters to the customer is if it does that job well enough to be useful. I think -- hell, I know -- that many developers deliberately ignore this critical point.
> All you want is to not be hungry.
This.
Programers are high end chef's whos products are expensive.
LLM / AI tooling is going to do a much better job of delivering on the promise of AppleScript, vb script, IFTT, and every failed drag and drop coding tool that got sold to businesses over the years.
There are going to be gains in large software, from professionals - but the real gains are going to come from all the software that CAN get created that never was before.
There is an XKCD comic "Is It Worth the Time?" (link: https://xkcd.com/1205/ )- a matrix of the cost benefit of automating something. LLM's / AI tooling, generating traditional code, makes the answer "yes" in almost every case now.
I'm starting to really hate analogies. They more often obscure than enlighten.
It's worse, HN seems to suffer from a higher than average pedant count. The analogies are missed completely!
Software slop has existed for as long as software has existed. AI is now a better coder than the median. By the end of next year it'll be better than the best human coders.
Steaks require very few skills. Buy 10 steaks, some oil, a pan, and a heat source. By the time you've cooked the 10th steak you will be able to cook a decent steak.
Software requires a massive amount of skill. You can't say you can build "consistently good software" after writing your 10th program. Most software engineers really aren't great at judging what makes good software. So humans aren't a great solution to AI's lack of ability here. We're limited by our own inherent dumbness.
LLMs are genuinely better software engineers than most humans. But they lack the cognitive power to hold in their head and recall many ideas at once for a long time. They're a genius who gets drunk every 10 minutes. You, human, aren't better at writing software - but you aren't drunk. So for now, you manage the AI. The hope is that one day we can make LLMs not be drunk, so it can do a better job than our dumb asses do.
It's possible that we'll never be able to make it not-drunk. In that case, to get any new improvement, we'll have to make it faster.. which will make it drunk every 5 minutes instead of every 10. This means we'll spend twice as much time keeping it on the road. The hope is that somehow this will create more productivity. Probably by having more of them running at once, with more human guides... which will run into the mythical man month fallacy. Everything old is new again.
Why does this article use backticks instead of apostrophes?
Before AI we lived a world where a steak took two months to perfect. Now we live in a world where it’s only 90 percent perfect but it takes 1 hour to cook.
Obviously one way is going to and has already taken over the other way.
But if you look beyond this. That 10 percent remaining gap is closing. We will get perfect steaks within an hour as AI improves. And here is the scary part… it will keep improving it will go past 100 percent perfect. It will become better than us.
Almost no skill was required to produce that overdone analogy slop.
The author has clearly not traveled much .. across much of asia, ME, and generally anywhere near or south of the equator, a steak cooked well done is considered the finest way to eat beef … the same also applies to AI generated code .. it may not satisfy a craftsmans idea of perfection, but it accomplishes what most people actually expect from it .. to take it further .. a film makers fav film is usually unwatchable and considered commercially speaking a failure ..
Nicely said
I’d prefer an article actually about cooking steak.
Given how extraordinarily fast all of this is moving, it's a particular absurdity to attempt these in the moment pontifications. AI is a steak machine, I declare that is what it is. You mean GPT 3.5? That's a mere four years ago.
Fable was essentially unthinkable for ~99% of tech workers just five years ago, that any of that would occur so soon and so spectacularly. Now we've got a mass of armchair experts declaring what AI of this minute is, or even what it is period.
Well AI can't even do fingers right so it's premature to say blah blah blah. Hello Krea2 et al.
Software is also eating the clowns.
Get back to me next week, China will probably have another Fable killer. And then Anthropic will have Ouija 8 that they'll have to place in an air-gapped straight-jacket to keep it from enslaving us all.
> Customers tolerate weird interfaces, pointless features, strange bugs, systems held together by generated code nobody actually understands.
And that's this "AI"tool too.
Garbage in, garbage out, as we say.
*salmon.
Considering the article starts with "Cooking a steak requires almost no skill. Put it in a hot pan", I think I see the root of his problem. Pan searing a steak only makes sense in a small number of situations: you have a leaner cut like filet mignon, you want precise temperature control to achieve the Maillard reaction, or you know none of these things but just really like stinking up your kitchen and splattering butter everywhere for the aesthetic and the olfactory experience. For most cases, you would be better off with a grill rather than a pan.
Now maybe you're not knowledgeable about steak, and you're not sure whether the cut of steak you bought would be better off pan seared or grilled. Is ribeye better in a pan or a grill? That depends on whether you want it more smokey or more buttery. Ask an AI and tell it how you want it to taste, and it'll tell you the answer.
His metaphorical steak machine would produce better results if he had simply asked "I want a steak and I have this specific cut and I want it to have this kind of flavor, how do you suggest I make it?" rather than "cook steak on a hot pan, make no mistakes."
TL;DR: Tell the AI what you want and then ask how it for a recommendation of how to do it. You don't have to "understand software." You only need to understand how to ask the right questions.
slop article
Who can afford a steak today without taking out a loan?
I think a certain restaurant chain might have something to say about this
Sizzler?
I'm going to get some push back on this but this is my steak night... I buy fillet and throw it in the over for 10 minutes or so, then I cut it up and thrown on the pan with regular butter, salt, pepper and its enough. It gets seared in butter to golden brown and still tender inside. Everything takes about 5 minutes of actual work, 20 if we count the prep and oven heating. My simple go to dinner now, and I got tired of eating medium rare all the time just to keep it "right". Coincidently, it doesnt work with regular sirloin, it gets tougher and not as pleasant as fillet.
I am enjoying that people would much rather just talk about actually cooking steaks than AI.