The central point, that working with agents is some tried and true variation of existing processes, is incorrect.
Managing agents has some similarities with EM and program management but a whole lot of other dimensions like token use, avoiding drift, successful concurrency at scale, variations in prompting, testing, evaluation, etc., not to mention that the agents are hyperintelligent coders with zero common sense and a penchant for extremely literal interpretation and ultra-verbosity.
This just reads like engineers. Token use is time, labor cost. Drift happens with humans too when they’re over-extended on a task with not enough check in. That’s why we have standup, to avoid drift. Prompting is management. Zero common sense and a penchant for extremely literal interpretation? That sounds like an average engineer who may or may not be on the spectrum.
Ultra-verbosity, okay you got me, humans don’t do that. They do over complexity though, when it makes them feel smart.
This isn't an issue for engineers only, it is basically ingrained in the current system in the world. The problem is in simple terms; the fact that being a polymath is absurdly hard in today's world.
You can't not reinvent something if you don't know it exists in the first place.
The thing about software is that it slots nicely into every other field, making it a _very_ good base to work off of to get concepts (aka the modern bootleg polymath) that have likely been invented in other fields (with different names).
I'm a hands on engineering leader for a team of about 20 engineers and I've been spending the past 6 months trying to get my team to understand just this.
On Friday we had a coffee hour to share how we've been working recently and they all seemed perplexed at the workflows I've been adopting. It seems natural to me as someone who's been a manager for some time now, but very alien to all those who've never gone down that path.
That's not to say my workflows are superior, but they're extremely different to some of my teams now. In reality it's just leaning heavily on things like prds, limiting communication between agents, etc.
Out of curiosity, what was the path that the engineers have been taking? The limiting communication between agents to my mind is obvious to anyone who hasn't swallowed orthodoxy Agile development completely whole. And in my experience the people who have, are rarely the devs. Context switching, between agents, people or whoever, require ramp up time to relearn context. It's always slower and more expensive, other potential upsides about long term training or developers being more replaceable notwithstanding. But no agent gets long term training, not in a 1M context window.
And our in house "Agile Practitioners" (their job title) pushed this to a state where the next four tickets in a task get broken down to four hour long pieces of work go to four different people, regardless of their context or skill set rather than what two years ago was what one person just owned and did in a day. Work in progress is minimized by the metrics, but it doubled the overall man hours. And the Ops Director gets promoted on that WIP metric while half the team quit citing the changes due to the added communication burden and lack of ownership.
I'm thinking some people are approaching agentic development in the same manner, splitting work up across agents too aggressively, giving each agent new context/upfront plans, when the one session could have done the whole problem in the one context window.
Right, the ultimate context switch is constant pinging and meetings. There’s very little reason to have engineers be in 5+ meetings a day. But at a lot of orgs, they are, and that’s an extreme context switch cost 5x a day. It’s different if it’s a purely technical meeting about the problem you’re solving right now, but they’re typically not.
In the few instances that I’ve been able to see a total stranger prompt a model, I learn a lot about them.
The model output is fairly uniform because the models are fairly uniform, but the input is how I understand the prompter’s theory of mind for the LLM.
The higher fidelity the theory of mind, the more productive the resulting conversation is. Basic things, like knowing what the model is even aware of.
It’s okay at subjective product judgements with limited context, and so the best engineers make the most important decisions themselves, constraining the model’s solution space to something looking like success.
Maintaining consistent progress towards a common goal with a bunch of different perspectives is the job.
Relatedly, I fully expect the software business to fail to learn from (nearly) literally every other industry how to operate when your marginal costs are no longer zero. It drives me nuts the number of people in software who think that people in other industries are slow because they’re just not as smart as us, rather than because when it takes months to get a cast part made, it better be right the first time.
It's even funnier when you realize that not too long ago a decent chunk of the software industry did have to deal with the realities of having to press CDs, print boxes and ship them to retailers in time for the holiday season. The idea that you can ship a half-baked product directly to your end users, charge for it and promise it will get better over time (but only if you manage to secure enough VC funding in the meantime) is a relatively new invention.
I remember working in the games industry in the 90s and early 2000s; getting a game past Sony testing was so fricking difficult, but it was because once that black disc was cut, there was no way to get an update. You had to be on the money, first release.
Yet this is part correct, part wrong. Indeed every industry has some unique challenges that are not directly solvable. However, while a cast is absurdly expensive, the problem is being solved in large part by software in the FDM (3D printing) space and it's changing how things are built.
edit: this doesn't invalidate the OP thesis, which I strongly agree with. Except that it's not just engineers, this is a more general thing across many fields.
Do you work in this field or are you just reading that? I’ve been out of the game for over five years now, but as far as I can tell from people I know still in that world, if you’re going to make more than like … 3 of a part, casting is going to be much cheaper. And while you can do some prototyping on 3d prints, the nature of the material winds up different from a cast part.
I'm not building molds, but I'm actively researching the field in conjunction with my work. Based on what I'm seeing and reading things have changed in the last few years dramatically. Concretely in the selective laser sintering (SLS) Formlabs is redefining small-scale manufacturing. But don't get distracted by the "small-scale". It's a bit like how small servers replaced mainframes. Production lines are getting replaced by much more versatile alternatives.
Formlabs recently launched a $80K industrial-grade machine that works better than it's $500K alternatives.
So is injection molding more effective than additive manufacturing? This question is a bit like "no more interesting than the question of whether a submarine can swim".
You'd need hundreds, if not thousands, of parts to justify a mold for casting. And even then, by the time you got the mold made you could have already printed hundreds of parts on a print farm, so you're losing time too.
That's totally untrue. You clearly do not have any experience designing, building, manufacturing, testing, and certifying either cast metal or PBF parts. BTW, PBF is powerbed fusion and it is how you make 3D printed metal parts. FDM is a plastic manufacturing process and has nothing at all to do with replacing metal castings.
For a certain aircraft part that I designed, that measured approximately 16"x8"x10", if you needed up to 12 of them it was cheapest to 3D print them using powerbed fusion from AlSi10Mg, but with a 25% weight penalty due to the lower mechanical properties of 3D printed aluminum compared to cast. If you needed between 12 and 50 of them, it was cheapest to use 3D printing to make a wax preform then use the preform to make a traditional metal investment casting from A356. For more than 50 of them, it was cheapest to invest in permanent closed die tooling.
Other than for prototyping and extremely low volume production, additive has been a total flop for real-world manufacturing, and has not at all lived up to any of it's lofty promises.
I wrote FDM instead of additive manufacturing. I'm referring to is actually in SLS (selective laser sintering) space (which is what you refer as PBF?)
What I'm seeing is that companies like Formlabs (plastics) and divergent (metal) are changing the game and you have car parts to car engines to rocket engines being 3d printed.
See my other comment above for more links, but things seem to be changing fast in manufacturing.
Edit:
I think this quote sums the point though:
“With the increased throughput and decreased costs, Fuse X1 has completely changed our perspective on what kinds of projects our lab can support versus what we would have traditionally moved to an injection mold.”
Cody Jepsen, Engineering Technician, Additive Manufacturing at Tesla Giga NV
Do you see that changing as parts are designed for additive manufacturing? Things that basically cannot be made another way, but have strength, weight, and even functional benefits over a traditionally made part?
I’m thinking of say cooling channels built into walls of rocket combustion chambers, or car brake/axle assemblies, and some of the „organic“ designs that are lighter and stronger than a cast/machined part.
Yes, the future of software as an institution, if there is any, is driven by the loudest voices. Those loudest voices are thus driven by selected personalities as opposed to any accomplishment or title. The people who are the highest achievers are rarely the loudest voices as they tend to be the people spending time solving real problems as opposed to the people who just talk about themselves, the forest for the trees.
IDK, I think it's been a bit of a (reverse?) bell curve of sorts.
30+ years ago, You had to be a -lot- more careful about how you handled releasing software. Your stuff was in boxes on shelves, most people didn't have internet access to 'download an update' so fixing anything had a long tail support cost (i.e. paying for the media containing the fix and shipping).
But there's a fine line. As an example, If your shop is having a vendor do brand new stuff, -compartmentalized from your other stuff-, using technologies that are actually out of support vs an in-support version is a free/zero effort update versus using the outdated stuff... Something went wrong on the management side.
How many updates did one apply to Windows 3.0 and Word in those days??
Sure, there might have been a “hotfix” available for download over a long distance BBS, but virtually nobody would have ever done so (even if they were lucky enough to have a modem). Maybe large companies applied such when impacted.
Even in the Windows XP days,
I think windowsupdate.com was still a manually triggered process…
For a while, bundled drivers were a steaming pile of crap in quality and several generations of Windows suffered greatly until Microsoft got vendors in line and updates became common.
I remember in 1999 reading some tech executive (with tech education, I guess) complain, like "I'd like to make executive financial reporting, but not like this horrendous and too complicated accounting". Then he proceeded with ideas. Well, the thing is that all our terms we use are lines from summarized accounting balance sheet, from either assets side, or liabilities. And if you detach them from the rest, the list of figures won't make any sense. I learned that studying accounting in the uni.
Also what I learned back then was that accounting was invented for the same purpose the blockchain: to make fraud in the workplace very hard.
I remember engineers reinvent geospatial technologies, avoiding such thing as map projections, and in the end still came to them.
A few years ago a software engineer asked a few Chemical/Civil/Mechanical engineers and the majority of "real" engineers seemed to think we were [1]. I'm not aware of a better study but happy to be proven wrong, since I generally refer to myself as a Software Developer
Not sure it’s obvious reasons, but understandable, sure. When someone questions your credentials the default is to tell them to jump, whether that’s a good idea or not.
Because Engineering is an incredibly rigorous discipline and title that requires so much work to obtain. It’s like a massage therapist calling themself a doctor. Computer Science isn’t Engineering.
Sure but out of my engineer friends I've never heard of them sneering a programmer calling themselves and engineer, they literally don't care and just worry about their own job. It seems only programmers seem to care about their lack of a proper engineer title.
Absolutely. Engineering is defined by it learning from history, it's an art build up from thousands of years of human attempts to alter the world around them. In a modern sense engineers inevitably require a certain amount of schooling, certification, and most critically of all professional and ethical standards.
Do people seriously care about this debate? I don't even use the term software engineer. I usually just say [computer] programmer. Though depending on my mood I might say I am virtual firefighter, and believe me, there are always virtual fires that need to be put out.
>You can make lots of money by making something appear novel and undiscovered, and consequently make yourself sound smart and cutting edge. Nobody raises a round to apply a well understood discipline correctly.
I'm curious about this. I thought investors preferred safe bets?
On the other hand, I know that if you're too early, it can be impossible to make a business work (or even to pitch the idea in the first place).
It's really funny, I've read this essay before a long time ago and it's really interesting playing back the many times both in realising it in retrospect and having seen it playing out before my eyes, that the last part really comes to the fore
> One final point: I expect none of you to really get what I’m talking about here, because this principle also applies to itself. But I fully expect I’ll get the occasional email saying “Oh! so that’s what you meant.” or “Why didn’t you tell me that?” I did, but you can’t tell people anything.
In my youthful naivety I thought I understood this, but now years later I can more deeply appreciate the message that you really can't tell people anything
He cites Brooks as proving that communication friction is quadratic in headcount. The exponent does not need to be exactly, or even nearly, 2; 1 + epsilon is already fatal. This is very closely related to Coase's ceiling.
That's a strawman argument, because most engineers aren't engineers. The vast majority of engineers have no consequences for their shite. Neither is certification a panacea against bad engineers.
Watch some electronic engineers design circuits that don't work.
Watch some geotech engineers make up overspecified bullshit: because they're paid per hour and their work is often just a glorified tickbox (where they have no real consequences for most of their failure risk).
Even with egregious design failures by engineers, they often get away with it for a variety of reasons. A building collapsed due to the Christchurch earthquakes - failures by different engineers with little harm to them.
That is categorically untrue. Professional Engineers not only face consequences, they are personally liable for the work they put their stamp on, and they can lose their license
The Christchurch CTV building was supposedly built to the code standard of the early 1980s, it was a non-ductile reinforced concrete structure.
Supposedly is heavily stressed, the Royal Commission investigating found a lack of communication between the official structural design engineers (Alan Reay Consultants), the actual engineer employed for the design (David Harding) and somewhat more critically the "engineer" on the ground implementing the building's construction (Gerald Shirtcliff).
Gerald Shirtcliff had faked his engineering degree.
Despite all this the building passed inspection and stood for 25 years, passing two additional inspections in 2010 (a year prior to collapse) after two separate earthquakes.
> they often get away with it for a variety of reasons.
The documents show police wanted to charge Harding and Reay with manslaughter. Christchurch Crown Solicitor Mark Zarifeh believed the evidential sufficiency test was met by expert evidence from engineering firm Beca.
Deputy Solicitor-General Brendan Horsley disagreed.
Horsley said for charges of negligent manslaughter, the Crown had to prove the conduct was so bad it deserved to be condemned as a serious crime. It would be difficult to show the conduct was one of the main causes of the deaths.
"A key difficulty for the prosecution would be in proving the CTV building would not have collapsed in the absence of the identified design errors," he said.
Another obstacle was the length of time between the design and when the deaths occurred.
~ stuff.co.nz
That's the dot point overview, and FWiW the actual design engineers got professional black marks on their careers and the case is still heavily studied in formal civil engineering courses.
This is mostly wrong as far as motivation is concerned. Some people just by default figure stuff out with what they know. Others accumulate knowledge. If you go too far in either direction you are probably going to struggle, but it's just different people having different nature.
The modern software engineer throws away the requirements specification because the computers he commands are fast and the people he negotiates with are slow. He ignores archival science because his business records are digital and the musings of a geriatric archivist are not compatible with PostgreSQL 18. He adopts the posture of an industry thought leader on X but has never read a standards document in his life. He believes he has little in common with the Roman aqueduct engineer but finds himself in management meetings pleading for more byte-sized stones for the stone god.
Lawyers are informed by ancient case law and bankers leverage trade instruments with roots in medieval Italy. Why is the software profession so exceptional that replacing hundreds of years of engineering wisdom with blog posts on agentic workflows is considered best practice?
True. Which is why I love this quote from the most beautiful works of art I have ever enjoyed, Clair Obscur: Expedition 33:
> You speak to me of perfection, but is there really any such thing? Or is it a siren, leading us astray. I happily predict that neither of us will ever summit that elusive peak, but I care little, as long as we walk together. Art gives us the perception of control. For a moment, as I paint, I find order among the chaos. That is, for me, a moment of pure contentment. And that is truly better than perfection.
Can we please stop with this endless, seemingly deliberate, misunderstanding of what Engineering actually involves?
You go to a 4 year Engineering school, get an actual Engineering degree. Then you apprentice with an Engineering company, and study for the state's license examination, one of the hardest tests you'll ever take. If you pass, then and only then are you an Engineer.
Real engineering is nothing like the slop portrayed in this article.
Engineering schools showcase failures of the past as a teaching tool. I still remember learning about the bridge that shook itself apart during my freshman year. Engineers value safety margins and failure analysis.
Nobody gets rewarded for rejecting the newfangled thing. Even when the newfangled thing should be rejected. Same as scientific papers are very biased against reporting nulls.
There’s a good list of reasons why we do this, but it leaves out the biggest one, at least for me: building stuff is fun. Reinventing stuff is fun. It’s the most natural hammer to reach for whenever I encounter a nail. Not necessarily the best, but such is life.
I try to explain this on every post that complains about software reinventors. Why should people from the 60s and 70s get to have all the fun? So far my efforts have been in vain.
Man I love and agree with nearly everything about this post except that I wouldn't take the advice to follow waterfall or PMBOK literally.
The post laments that it's hard to find Dr. Royce's original waterfall paper. That's probably true, I have my copy from a compilation book “Ideas that Created the Future: Classic Papers of Computer Science” edited by Lewis [1].
I do agree that it's important to do things like scope out your demands of your AI agent, check-in on progress, give as clear a requirement and test cases as you can. But you've always been able to do that with agile methods, and LLMs are fast enough that you don't need to go full waterfall (and if anything it would be counter productive).
Dr. Royce's paper talks about literally thousands of pages of documentation being needed for any reasonably useful system. Good luck fitting that into even a 1M context window :P.
But the main point to thesis, that you can't just let your coders loose to do whatever and expect the right results even pre-dates Fred Brooks. I'd argue it goes all the way back to the beginning, to the comments about Baggage's computing machine where British politicians asked if it would generate the correct answers even with incorrect inputs.
The answer then is the same answer today: of course not, and expecting anything different is foolishness.
Unless I am mistaken, buses do not descend in a lift and travel underground. It's pure fantasy of course, but no we didn't somehow forget buses exist and reinvent bus stops.
This I can agree with:
- People can "rediscover" knowledge that was already established because they don't investigate other disciplines first.
- Agent engineering can benefit from decades of studies in management, requirements engineering, operations and process research.
But the article has a big irony: the article accuses engineers of oversimplifying other disciplines and then does exactly that itself.
> "Data science was statistics with a cooler name"
The author mentions David Donoho's "50 Years of Data Science" but that work presents a much more complicated concept of data science: one with data exploration, transformation, computing, visualization, modeling, etc. Even the study of data-analysis practice itself.
Look at that XKCD he posted with it.
Horn took a nuance argument and converted it into the "human slop" formula mocked in the XKCD:
complicated subject = simple thing I already understand
<new complicated thing> is really just <old familiar thing>
The cartoon is criticizing the same operation the article is doing.
Maybe that is the joke?
As someone that has studied engineering, designing a solution is very different than managing the resources around the design and implementation of that solution. Books like The Mythical Man Month and Making Things Happen deals with the management side, (from a team level) and The Pragmatic Programmer is from a personal level.
But TFA is ignoring the actual engineering side where you have to grapple with some primitives and assemble them in a way that do something valuable. And so in a way that is cost effective. That part is always answered by hand waves.
I don’t quite understand the management analogy. Agents aren’t people, they’re code generation machines. Part of managing is delegating and trusting the output of your employees without having to verify everything yourself. But if you’re “managing” an agent, how does that work? You can’t exactly hold an LLM responsible if it fucks something up.
Or are people really just yoloing and not even verifying that the code generation is correct? I know it’s a bit of a meme, but are people actually doing the meme in irl where there are actual consequences??
The author missed one key motivation: giant fucking ego. I don’t know why every damn software engineer thinks they are gods gift and can derive anything from first principles. The OP hinted at it with the physicist XKCD but I think it is far more dominant in software because the payouts for mediocrity are just astronomical.
Another obnoxious behaviour I’ve observed lately, is trying to attach whatever pre-existing pet methodologies one had to the AI-hype bandwagon like some sort of personal vindication orgy.
> You can make lots of money by making something appear novel and undiscovered, and consequently make yourself sound smart and cutting edge.
This is pretty much the crux of it. It's very similar to the strategy of undercutting a market with VC subsidies until it dies and can be replaced.
VC subsidies are no longer enough, now we need IPOs to keep it going...
An IPO and subsequent offers is hardly a temporary subsidy. That can keep going until the economy dies.
Mate, its so fucked, people are going to cry way worse than 2008.
people in these fields we pre-crying, building the torment nexus, because if they dont, they'll be in the "permanent underclass".
FOMO is going to kill more americans than COVID.
As an MCP engineer, how dare you
> MCP Engineer
LOL gotta put that on my resume
Sounds like Apple.
The central point, that working with agents is some tried and true variation of existing processes, is incorrect.
Managing agents has some similarities with EM and program management but a whole lot of other dimensions like token use, avoiding drift, successful concurrency at scale, variations in prompting, testing, evaluation, etc., not to mention that the agents are hyperintelligent coders with zero common sense and a penchant for extremely literal interpretation and ultra-verbosity.
This just reads like engineers. Token use is time, labor cost. Drift happens with humans too when they’re over-extended on a task with not enough check in. That’s why we have standup, to avoid drift. Prompting is management. Zero common sense and a penchant for extremely literal interpretation? That sounds like an average engineer who may or may not be on the spectrum.
Ultra-verbosity, okay you got me, humans don’t do that. They do over complexity though, when it makes them feel smart.
Yeah. Just because you can list a lot of ways that two things are similar, it doesn’t make them actually that similar in practice.
That’s why predicting the future is really really hard.
How many of these complications will anyone even remember in 12 months?
This isn't an issue for engineers only, it is basically ingrained in the current system in the world. The problem is in simple terms; the fact that being a polymath is absurdly hard in today's world.
You can't not reinvent something if you don't know it exists in the first place.
The thing about software is that it slots nicely into every other field, making it a _very_ good base to work off of to get concepts (aka the modern bootleg polymath) that have likely been invented in other fields (with different names).
I'm a hands on engineering leader for a team of about 20 engineers and I've been spending the past 6 months trying to get my team to understand just this.
On Friday we had a coffee hour to share how we've been working recently and they all seemed perplexed at the workflows I've been adopting. It seems natural to me as someone who's been a manager for some time now, but very alien to all those who've never gone down that path.
That's not to say my workflows are superior, but they're extremely different to some of my teams now. In reality it's just leaning heavily on things like prds, limiting communication between agents, etc.
Out of curiosity, what was the path that the engineers have been taking? The limiting communication between agents to my mind is obvious to anyone who hasn't swallowed orthodoxy Agile development completely whole. And in my experience the people who have, are rarely the devs. Context switching, between agents, people or whoever, require ramp up time to relearn context. It's always slower and more expensive, other potential upsides about long term training or developers being more replaceable notwithstanding. But no agent gets long term training, not in a 1M context window.
> obvious to anyone who hasn't swallowed orthodoxy Agile development completely whole.
What? One of the first principles in doing Lean and Agile correctly is limiting Work in Progress and avoiding context switching whenever possible.
And our in house "Agile Practitioners" (their job title) pushed this to a state where the next four tickets in a task get broken down to four hour long pieces of work go to four different people, regardless of their context or skill set rather than what two years ago was what one person just owned and did in a day. Work in progress is minimized by the metrics, but it doubled the overall man hours. And the Ops Director gets promoted on that WIP metric while half the team quit citing the changes due to the added communication burden and lack of ownership.
I'm thinking some people are approaching agentic development in the same manner, splitting work up across agents too aggressively, giving each agent new context/upfront plans, when the one session could have done the whole problem in the one context window.
Most organizations are not doing Agile correctly.
Everywhere I've worked claims to be doing Agile, but none of them have ever done anything to avoid context switching or limiting work in progress
Right, the ultimate context switch is constant pinging and meetings. There’s very little reason to have engineers be in 5+ meetings a day. But at a lot of orgs, they are, and that’s an extreme context switch cost 5x a day. It’s different if it’s a purely technical meeting about the problem you’re solving right now, but they’re typically not.
if its like anything i've experienced in multiple places, they are just implementing scrum and calling it a day.
meanwhile the business runs on quarterly projects that have to be done by a certain marketing date, so its usually scrum + waterfall over and over
In the few instances that I’ve been able to see a total stranger prompt a model, I learn a lot about them.
The model output is fairly uniform because the models are fairly uniform, but the input is how I understand the prompter’s theory of mind for the LLM.
The higher fidelity the theory of mind, the more productive the resulting conversation is. Basic things, like knowing what the model is even aware of.
It’s okay at subjective product judgements with limited context, and so the best engineers make the most important decisions themselves, constraining the model’s solution space to something looking like success.
Maintaining consistent progress towards a common goal with a bunch of different perspectives is the job.
Can you share some details of the workflows that work for you?
I would also love to hear more details.
Is anyone on your team not using multi agent?
Relatedly, I fully expect the software business to fail to learn from (nearly) literally every other industry how to operate when your marginal costs are no longer zero. It drives me nuts the number of people in software who think that people in other industries are slow because they’re just not as smart as us, rather than because when it takes months to get a cast part made, it better be right the first time.
It's even funnier when you realize that not too long ago a decent chunk of the software industry did have to deal with the realities of having to press CDs, print boxes and ship them to retailers in time for the holiday season. The idea that you can ship a half-baked product directly to your end users, charge for it and promise it will get better over time (but only if you manage to secure enough VC funding in the meantime) is a relatively new invention.
I remember working in the games industry in the 90s and early 2000s; getting a game past Sony testing was so fricking difficult, but it was because once that black disc was cut, there was no way to get an update. You had to be on the money, first release.
> did have to deal with the realities of having to press CDs, print boxes and ship them to retailers in time
This is because the we have these periodic influxes of scammers into the IT the industry, with their empty promises and trust me bro culture.
Yet this is part correct, part wrong. Indeed every industry has some unique challenges that are not directly solvable. However, while a cast is absurdly expensive, the problem is being solved in large part by software in the FDM (3D printing) space and it's changing how things are built.
edit: this doesn't invalidate the OP thesis, which I strongly agree with. Except that it's not just engineers, this is a more general thing across many fields.
Do you work in this field or are you just reading that? I’ve been out of the game for over five years now, but as far as I can tell from people I know still in that world, if you’re going to make more than like … 3 of a part, casting is going to be much cheaper. And while you can do some prototyping on 3d prints, the nature of the material winds up different from a cast part.
I'm not building molds, but I'm actively researching the field in conjunction with my work. Based on what I'm seeing and reading things have changed in the last few years dramatically. Concretely in the selective laser sintering (SLS) Formlabs is redefining small-scale manufacturing. But don't get distracted by the "small-scale". It's a bit like how small servers replaced mainframes. Production lines are getting replaced by much more versatile alternatives.
Formlabs recently launched a $80K industrial-grade machine that works better than it's $500K alternatives.
https://formlabs.com/blog/announcing-fuse-x1-industrial-sls/ https://formlabs.com/blog/diy-injection-molding/ https://www.youtube.com/watch?v=0TvjUVdn3YQ&t=2586s
Formlabs is used by Tesla to print car parts to Ukraine to print drones.
The thing about SLS is that it works with metal too. So you have startups like Divergent (recently raised $290M) in LA printing car engines https://www.latimes.com/b2b/space-tech/story/2026-06-23/dive...
Note that rocket engines are 3d printed too from SpaceX to India's Sky Roots https://www.selfcad.com/blog/how-spacex-uses-additive-manufa... https://3dprintingindustry.com/news/skyroots-vikram-1-reache...
So is injection molding more effective than additive manufacturing? This question is a bit like "no more interesting than the question of whether a submarine can swim".
But while you have metal at industrial scale you have "desktop" metal printers for ~$10K https://all3dp.com/4/first-look-at-the-scrap-1-desktop-metal...
You'd need hundreds, if not thousands, of parts to justify a mold for casting. And even then, by the time you got the mold made you could have already printed hundreds of parts on a print farm, so you're losing time too.
That's totally untrue. You clearly do not have any experience designing, building, manufacturing, testing, and certifying either cast metal or PBF parts. BTW, PBF is powerbed fusion and it is how you make 3D printed metal parts. FDM is a plastic manufacturing process and has nothing at all to do with replacing metal castings.
For a certain aircraft part that I designed, that measured approximately 16"x8"x10", if you needed up to 12 of them it was cheapest to 3D print them using powerbed fusion from AlSi10Mg, but with a 25% weight penalty due to the lower mechanical properties of 3D printed aluminum compared to cast. If you needed between 12 and 50 of them, it was cheapest to use 3D printing to make a wax preform then use the preform to make a traditional metal investment casting from A356. For more than 50 of them, it was cheapest to invest in permanent closed die tooling.
Other than for prototyping and extremely low volume production, additive has been a total flop for real-world manufacturing, and has not at all lived up to any of it's lofty promises.
I wrote FDM instead of additive manufacturing. I'm referring to is actually in SLS (selective laser sintering) space (which is what you refer as PBF?)
What I'm seeing is that companies like Formlabs (plastics) and divergent (metal) are changing the game and you have car parts to car engines to rocket engines being 3d printed.
See my other comment above for more links, but things seem to be changing fast in manufacturing.
Edit:
I think this quote sums the point though:
“With the increased throughput and decreased costs, Fuse X1 has completely changed our perspective on what kinds of projects our lab can support versus what we would have traditionally moved to an injection mold.” Cody Jepsen, Engineering Technician, Additive Manufacturing at Tesla Giga NV
Do you see that changing as parts are designed for additive manufacturing? Things that basically cannot be made another way, but have strength, weight, and even functional benefits over a traditionally made part?
I’m thinking of say cooling channels built into walls of rocket combustion chambers, or car brake/axle assemblies, and some of the „organic“ designs that are lighter and stronger than a cast/machined part.
Yes, the future of software as an institution, if there is any, is driven by the loudest voices. Those loudest voices are thus driven by selected personalities as opposed to any accomplishment or title. The people who are the highest achievers are rarely the loudest voices as they tend to be the people spending time solving real problems as opposed to the people who just talk about themselves, the forest for the trees.
As a buddy in construction used to say, concrete erasers are expensive.
Speaking of which, are you done with my cat5-stretcher?
IDK, I think it's been a bit of a (reverse?) bell curve of sorts.
30+ years ago, You had to be a -lot- more careful about how you handled releasing software. Your stuff was in boxes on shelves, most people didn't have internet access to 'download an update' so fixing anything had a long tail support cost (i.e. paying for the media containing the fix and shipping).
But there's a fine line. As an example, If your shop is having a vendor do brand new stuff, -compartmentalized from your other stuff-, using technologies that are actually out of support vs an in-support version is a free/zero effort update versus using the outdated stuff... Something went wrong on the management side.
Exactly!
How many updates did one apply to Windows 3.0 and Word in those days??
Sure, there might have been a “hotfix” available for download over a long distance BBS, but virtually nobody would have ever done so (even if they were lucky enough to have a modem). Maybe large companies applied such when impacted.
Even in the Windows XP days, I think windowsupdate.com was still a manually triggered process…
For a while, bundled drivers were a steaming pile of crap in quality and several generations of Windows suffered greatly until Microsoft got vendors in line and updates became common.
> Your stuff was in boxes on shelves...
Have you ever heard of an SLA? This "used to be" story about software has never been true. If you can't meet the agreement, then or now, you're done.
Software has always been a lot more diverse than just consumer goods. Why is that your only evidence of how you think software projects are managed?
> It drives me nuts the number of people in software who think that people in other industries are slow because they’re just not as smart as us. . .
Ha! I studied computer science and became a programmer because I knew I was too dumb for med. school and the other engineering paths.
Don't discount yourself like that mate. Programming is much harder than the run of the mill medical work or engineering work.
I remember in 1999 reading some tech executive (with tech education, I guess) complain, like "I'd like to make executive financial reporting, but not like this horrendous and too complicated accounting". Then he proceeded with ideas. Well, the thing is that all our terms we use are lines from summarized accounting balance sheet, from either assets side, or liabilities. And if you detach them from the rest, the list of figures won't make any sense. I learned that studying accounting in the uni.
Also what I learned back then was that accounting was invented for the same purpose the blockchain: to make fraud in the workplace very hard.
I remember engineers reinvent geospatial technologies, avoiding such thing as map projections, and in the end still came to them.
Seems like this could be categorized as yet another reason why software developers are not engineers.
A few years ago a software engineer asked a few Chemical/Civil/Mechanical engineers and the majority of "real" engineers seemed to think we were [1]. I'm not aware of a better study but happy to be proven wrong, since I generally refer to myself as a Software Developer
[1] https://www.hillelwayne.com/post/are-we-really-engineers/
I happen to be both. Engineering is a mindset, and to call yourself an engineer, people should be at least this: https://www.internationalengineeringalliance.org/accords/was...
I never understood why actual engineers(eg. Civil) have not complained more that programmers/coders stole that title.
they have, loudly and repeatedly, it doesn't really make it very far on HN for obvious reasons.
Not sure it’s obvious reasons, but understandable, sure. When someone questions your credentials the default is to tell them to jump, whether that’s a good idea or not.
HN removes every post that doesn't get upvotes, especially if they get lots of comments.
Why would they care? And a lot of them turned to coding anyway.
Because Engineering is an incredibly rigorous discipline and title that requires so much work to obtain. It’s like a massage therapist calling themself a doctor. Computer Science isn’t Engineering.
Sure but out of my engineer friends I've never heard of them sneering a programmer calling themselves and engineer, they literally don't care and just worry about their own job. It seems only programmers seem to care about their lack of a proper engineer title.
Absolutely. Engineering is defined by it learning from history, it's an art build up from thousands of years of human attempts to alter the world around them. In a modern sense engineers inevitably require a certain amount of schooling, certification, and most critically of all professional and ethical standards.
Do people seriously care about this debate? I don't even use the term software engineer. I usually just say [computer] programmer. Though depending on my mood I might say I am virtual firefighter, and believe me, there are always virtual fires that need to be put out.
>You can make lots of money by making something appear novel and undiscovered, and consequently make yourself sound smart and cutting edge. Nobody raises a round to apply a well understood discipline correctly.
I'm curious about this. I thought investors preferred safe bets?
On the other hand, I know that if you're too early, it can be impossible to make a business work (or even to pitch the idea in the first place).
Related: You can't tell people anything (2004)
https://web.archive.org/web/20091025030730/https://habitatch...
It's really funny, I've read this essay before a long time ago and it's really interesting playing back the many times both in realising it in retrospect and having seen it playing out before my eyes, that the last part really comes to the fore
In my youthful naivety I thought I understood this, but now years later I can more deeply appreciate the message that you really can't tell people anythingHe cites Brooks as proving that communication friction is quadratic in headcount. The exponent does not need to be exactly, or even nearly, 2; 1 + epsilon is already fatal. This is very closely related to Coase's ceiling.
That’s because most Software Engineers aren’t Engineers, they’re computer science majors. It’s a completely different discipline.
Yep, you actually do get Electronic & Software Engineers, which has CS subjects, with an Engineering discipline.
That's a strawman argument, because most engineers aren't engineers. The vast majority of engineers have no consequences for their shite. Neither is certification a panacea against bad engineers.
Watch mechanical engineers design plastic rubbish.
Watch some electronic engineers design circuits that don't work.
Watch some geotech engineers make up overspecified bullshit: because they're paid per hour and their work is often just a glorified tickbox (where they have no real consequences for most of their failure risk).
Even with egregious design failures by engineers, they often get away with it for a variety of reasons. A building collapsed due to the Christchurch earthquakes - failures by different engineers with little harm to them.
Yes. This. Also I don't think most software engineers are CS majors, either :)
That is categorically untrue. Professional Engineers not only face consequences, they are personally liable for the work they put their stamp on, and they can lose their license
The Christchurch CTV building was supposedly built to the code standard of the early 1980s, it was a non-ductile reinforced concrete structure.
Supposedly is heavily stressed, the Royal Commission investigating found a lack of communication between the official structural design engineers (Alan Reay Consultants), the actual engineer employed for the design (David Harding) and somewhat more critically the "engineer" on the ground implementing the building's construction (Gerald Shirtcliff).
Gerald Shirtcliff had faked his engineering degree.
Despite all this the building passed inspection and stood for 25 years, passing two additional inspections in 2010 (a year prior to collapse) after two separate earthquakes.
> they often get away with it for a variety of reasons.
~ stuff.co.nzThat's the dot point overview, and FWiW the actual design engineers got professional black marks on their careers and the case is still heavily studied in formal civil engineering courses.
* https://en.wikipedia.org/wiki/CTV_Building
* https://www.stuff.co.nz/national/99399561/police-will-not-pr...
>The vast majority of engineers have no consequences for their shite.
What do you base this on? In my country, engineers are legally responsible for their designs.
Reminds me of a quote by my old prof at Oxford:
"We used to just call this stuff chemistry; but citations and funding didn't really take off until we started calling it nanotechnology."
Related: https://news.ycombinator.com/item?id=49309451
> Working with AI feels more like leadership than coding
"Suddenly, waterfall is the thing to do."
Matches my experience. Like 90% of the work is the spec.
Except instead of weeks researching its like a few hours talking with an agent.
This is mostly wrong as far as motivation is concerned. Some people just by default figure stuff out with what they know. Others accumulate knowledge. If you go too far in either direction you are probably going to struggle, but it's just different people having different nature.
The modern software engineer throws away the requirements specification because the computers he commands are fast and the people he negotiates with are slow. He ignores archival science because his business records are digital and the musings of a geriatric archivist are not compatible with PostgreSQL 18. He adopts the posture of an industry thought leader on X but has never read a standards document in his life. He believes he has little in common with the Roman aqueduct engineer but finds himself in management meetings pleading for more byte-sized stones for the stone god.
Lawyers are informed by ancient case law and bankers leverage trade instruments with roots in medieval Italy. Why is the software profession so exceptional that replacing hundreds of years of engineering wisdom with blog posts on agentic workflows is considered best practice?
A comments section about an HN comment about a post. We can recurse even deeper.
That's a conversation in multiple threads happening in blogs. Much better than social media.
He says "engineers" but he means code monkeys. Actual engineering is all about learning from past failures
This is why we do not have any real engineers at the helm of any nation, state, multi-national enterprise or anything really.
The current world conceptions of engineers are that they're the negative ones, the naysayers.
Aren't there a lot of engineers/scientists near the top of the CCP?
But you and I both know that we don't talk about that :)
All while creating new ones. :)
Nobody's perfect ¯\_(ツ)_/¯
True. Which is why I love this quote from the most beautiful works of art I have ever enjoyed, Clair Obscur: Expedition 33:
> You speak to me of perfection, but is there really any such thing? Or is it a siren, leading us astray. I happily predict that neither of us will ever summit that elusive peak, but I care little, as long as we walk together. Art gives us the perception of control. For a moment, as I paint, I find order among the chaos. That is, for me, a moment of pure contentment. And that is truly better than perfection.
Can we please stop with this endless, seemingly deliberate, misunderstanding of what Engineering actually involves?
You go to a 4 year Engineering school, get an actual Engineering degree. Then you apprentice with an Engineering company, and study for the state's license examination, one of the hardest tests you'll ever take. If you pass, then and only then are you an Engineer.
Real engineering is nothing like the slop portrayed in this article.
Engineering schools showcase failures of the past as a teaching tool. I still remember learning about the bridge that shook itself apart during my freshman year. Engineers value safety margins and failure analysis.
Engineers DO learn from history.
Most HUMANS will do anything to avoid learning from history. Engineers are merely an example. Sigh.
Take my money. This is better than "The Pragmatic Engineer".
More then just engineers, we all need to study history
Nobody gets rewarded for rejecting the newfangled thing. Even when the newfangled thing should be rejected. Same as scientific papers are very biased against reporting nulls.
There’s a good list of reasons why we do this, but it leaves out the biggest one, at least for me: building stuff is fun. Reinventing stuff is fun. It’s the most natural hammer to reach for whenever I encounter a nail. Not necessarily the best, but such is life.
I try to explain this on every post that complains about software reinventors. Why should people from the 60s and 70s get to have all the fun? So far my efforts have been in vain.
Finding a thing, using it, finding out it sucks, and then making your own version that's better and doesn't suck, is an entirely natural instinct.
It is, but we often skip over those first three parts and go straight to making our own version.
Man I love and agree with nearly everything about this post except that I wouldn't take the advice to follow waterfall or PMBOK literally.
The post laments that it's hard to find Dr. Royce's original waterfall paper. That's probably true, I have my copy from a compilation book “Ideas that Created the Future: Classic Papers of Computer Science” edited by Lewis [1].
I do agree that it's important to do things like scope out your demands of your AI agent, check-in on progress, give as clear a requirement and test cases as you can. But you've always been able to do that with agile methods, and LLMs are fast enough that you don't need to go full waterfall (and if anything it would be counter productive).
Dr. Royce's paper talks about literally thousands of pages of documentation being needed for any reasonably useful system. Good luck fitting that into even a 1M context window :P.
But the main point to thesis, that you can't just let your coders loose to do whatever and expect the right results even pre-dates Fred Brooks. I'd argue it goes all the way back to the beginning, to the comments about Baggage's computing machine where British politicians asked if it would generate the correct answers even with incorrect inputs.
The answer then is the same answer today: of course not, and expecting anything different is foolishness.
[1] https://www.amazon.com/dp/0262045303
Software developers aren't real engineers; such an ironic title.
food engineers are considered real engineers. There is no way food engineering is more engineering than software engineering.
The majority of modern software engineering doesn't have human illness and or death as a consequence of failure.
That's front and centre as a concern in Food, Chemical, Civil, Electrical, Mechanical etc. engineering disciplines.
Say that to the victims of social media.
Slightly off topic but...
> We even reinvented bus stops.
Unless I am mistaken, buses do not descend in a lift and travel underground. It's pure fantasy of course, but no we didn't somehow forget buses exist and reinvent bus stops.
Solo programming is like managing a team of developers that can only communicate and contribute in one temporal direction.
Sometimes two or more developers may appear to coexist, but it's actually a simulated concurrency achieved by rapid context switching
Every time you close a file to open another, you are setting up a different context, therefore you are different agents.
This I can agree with: - People can "rediscover" knowledge that was already established because they don't investigate other disciplines first. - Agent engineering can benefit from decades of studies in management, requirements engineering, operations and process research.
But the article has a big irony: the article accuses engineers of oversimplifying other disciplines and then does exactly that itself.
> "Data science was statistics with a cooler name"
The author mentions David Donoho's "50 Years of Data Science" but that work presents a much more complicated concept of data science: one with data exploration, transformation, computing, visualization, modeling, etc. Even the study of data-analysis practice itself.
Look at that XKCD he posted with it.
Horn took a nuance argument and converted it into the "human slop" formula mocked in the XKCD:
complicated subject = simple thing I already understand <new complicated thing> is really just <old familiar thing>
The cartoon is criticizing the same operation the article is doing. Maybe that is the joke?
As someone that has studied engineering, designing a solution is very different than managing the resources around the design and implementation of that solution. Books like The Mythical Man Month and Making Things Happen deals with the management side, (from a team level) and The Pragmatic Programmer is from a personal level.
But TFA is ignoring the actual engineering side where you have to grapple with some primitives and assemble them in a way that do something valuable. And so in a way that is cost effective. That part is always answered by hand waves.
I don’t quite understand the management analogy. Agents aren’t people, they’re code generation machines. Part of managing is delegating and trusting the output of your employees without having to verify everything yourself. But if you’re “managing” an agent, how does that work? You can’t exactly hold an LLM responsible if it fucks something up.
Or are people really just yoloing and not even verifying that the code generation is correct? I know it’s a bit of a meme, but are people actually doing the meme in irl where there are actual consequences??
See also: people.
The author missed one key motivation: giant fucking ego. I don’t know why every damn software engineer thinks they are gods gift and can derive anything from first principles. The OP hinted at it with the physicist XKCD but I think it is far more dominant in software because the payouts for mediocrity are just astronomical.
Another obnoxious behaviour I’ve observed lately, is trying to attach whatever pre-existing pet methodologies one had to the AI-hype bandwagon like some sort of personal vindication orgy.
We don’t have long before it is all made irrelevant so need to make one final dash for glory before it’s too late.