For things like "add this feature" you can verify by trying out the feature and seeing if it works. Of course, that won't tell you if there are edge-case bugs or security holes or other design flaws, but you can get to a positive confirmation of the happy case pretty quickly.
NP-Hard, even[1]!
[1] https://www.i-programmer.info/news/112-theory/3830-physics-i...
AI’s only part of this was helping review things and finding bugs. More bugs, bugs that would have been found later, whatever.
Postgres did not have a slop issue and it’s wrong for people to insinuate it.
“Find the issues in this code” vs “write this feature” are different problems.
The first one is easier because the feedback loop is llm to code/test and trivially scalable.
The second one is llm to human which is going to be slower because it can’t read your mind.
The tighter and more deterministic the loop the more llms excel.
make a bunch of "build the feature" vs "heres a detailed spec" prompts, and see which one does better.
if you take out the AI boogieman, and instead go with "next token prediction" then "people are doing it wrong" is much more reasonable
I've worked for CEOs and managers that would insist on ridiculous ideas. Like that time a CEO (let's call him Cedric) came in with a printed powerpoint slide that had a screenshot of the website with a "login form" drawn on top of it.
Cedric: "I need you guys to put this onto that scrunt board thing and make this before friday".
Me (tech lead at that time): "But what features would users need login for? Where or whom needs the profile data? We'd need to work this out in detail and plan it. Certainly not even this month".
Cedric: "I don't know, it's for you to figure out". and then "It can't be hard, it's just two squares one for a username one for a password" (he forgot about the submit-button, the third "square", he didn't even fathom the idea that something has to lie "behind" the "squares").
I recall this anecdote whenever I feel anxious about whether soon "stakeholders" can just chat to an AI agent and have it build features, rather than hire me to do so. Sure, a lot of them will be able to. But there's always Cedrics. Those who not only fail to communicate their ideas in a way that even an AI agent would understand, they lack any understanding of what tradeoffs of their requested features are.
The Cedric's would be happy with an AI agent that complies and starts churning away. An agent that blindly accepts any request and builds feature upon feature. Making their product worse in every sense: unusable, unmaintainable, debt-ridden, security-nightmares.
Their agent will build features that are plain illegal. Or features that users hate, or confuse them. It will oblige requests that drive users away, or changes parts that users/integrations rely on. Their AI agent will gladly add a "form" to extract all PII from their database, but on request, will remove the authorization from it. Because "you are absolutely right. If we don't link to the page from the menu, no-one can reach it".
I'm not making the last one up. Another "Cedric" alike recently hired me to look at his vibe-coded app because his go-to developer gave up and dropped the project. It had loads of such extractors, dashboards and overviews. Online. Unauthenticated. His claude, however, did add the pages to the sitemap, so a customer found their private data on google and contacted "Cedric", after which he asked me.
No, AI is getting smarter, but as long as the "Cedrics" remain as stupid as ever, I don't fear for my job.
AI merely makes certain folk stay at the start of the Dunning-Kruger curve for longer and longer. Having zero understanding of a system and only see the user-interface and concluding that is somehow all that is involved means people end up thinking everything is far, far simpler than it actually is.
And nothing (and I really do mean "nothing") is simple. We have layers and layers of complexity and idiocratic systems with complex interworking and are creating the opportunity to create things based merely on the appearance of the icing on the cake, or the output of a sausage machine without understanding anything about farming cows.
any kind of pattern matching / transformation is going to be an ideal task for LLMs
But personally I know AI can write absolutely insanely complex things in very short amount of time, that just work and work well. I'm running things I would not dream of just 1 year ago, like advanced ISP pipelines on (formerly) high end Kintex K480T based FPGA accelerators pretty much fully written by AI, from the RTL, to all the testbenches, algorithm research, calibration tooling, etc. etc. And that's just one weekend project.
And they also won't believe you since they formed their opinion 6 months ago when code quality was 80% slop. Their loss though I guess? At least they'll be more surprised than us when they actually do lose their jobs? Maybe blissful ignorance is better!
In my very recent experience, it's good for routine things, but it consistently chooses the wrong sync primitives and dynamic abstractions for performance / extensibility. It puts everything in a Mutex/RWLock, which is terrible for write-heavy / contested values, and it prefers enums over traits, which is not extensible and can get inefficient.
But we're rapidly heading into a world where people no longer understand the difference, and therefore everything looks fine to them.
If possible capturing golden inputs/outputs for a block of code, and just asking the agent to optimize while ensuring those goldens pass can be quite effective.
There's a lot of performance-sensitive code where I work (i.e. market data processing) and LLMs have been very effective at reducing our p50, p99, etc...
Occasionally I do spot things and get a "you're right, I was wrong". But it's got to the point that my thought is no longer "of course I am... shitty slop machine" but "phew, I'm not totally obsolete quite yet!".
Maybe the problem is that it’s called “Generative AI“. It might be to use it to generate critique of content as opposed to generate the content itself.
https://www.postgresql.org/message-id/20140818195634.2543.56...
It seems quite possible that someone wants to make it seem like LLMs are causing problems by over-reacting to things that could just be fixed, or delay the release slightly (or maybe waiting until late before a release to kill features they don't like?) Whatever it is, this seems like a dysfunctional process.
And instead of releasing substandard new features, they rather focus on getting rid of the backlog and have more time to test.
LLMs are in effect directly responsible for this (in a positive way by finding a ton of issues that can be security issues). Reality is with LLMs now in the world, more work needs to be done on existing software as their ability to find bugs/security weaknesses is strong.
When the old bugs/security weaknesses are fixed, things will go back to a more normal release schedule.
Are the features designed well (from a product perspective) but implemented poorly? Again, the process is dysfunctional to let them get this far before it notices.
Were there known issues with these that were planned to be fix, but the fixes didn't happen in time? That would make sense and it happens, but how is an LLM making that worse? LLMs in the worst case let you get the same work done more quickly, because you can always get the work done as quickly as you ever could by just not using an LLM. So the strongest thing you might say here is that you thought LLMs would let you get a feature into shape before it was too late, and for whatever reason that was overly optimistic.
> The long-term picture is less clear and in some ways more troubling: given that we have forty years of bugs that humans were unable to find but LLMs are quite capable of discovering, how are we ever going to catch up, and how will PostgreSQL continue to innovate? Right now, I’m not sure. What I do know is that I, and many others, are working really hard to figure it out.
Two thoughts:
1. If the signal quality is high--the bug reports are generally valid, the submitted fixes are generally viable (if needing some attention)--then this all seems great. Pause innovation for a while, you can afford it. Focus on burning down the bug count, and resume feature development when things are more stable.
2. If the signal quality isn't good, or is sufficiently mixed that a lot of time is being wasted on low quality contributions... that's more complex. I don't know what to do about that. Maybe stop accepting contributions from unknown contributors? No idea.
From the post I'm not sure whether they're experiencing a (1) or a (2). If it's a (1) that's really interesting, and gives me some glimmer of hope for the future.
[edit] this piece of shit website won't let me reply "you're posting too fast.." so I'll respond to the two replies here (gondalez and sverhagen had basically the same point):
Yes that's a very good point. Not every valid bugfix is equivalently urgent. That's another critical dimension of quality--not burdening maintainers with pedantic bugs. This kind of contextual nuance is, coincidentally, something AI is absolutely awful at.
this doesn't seem good attitude for critical software like database
[0] https://www.postgresql.org/message-id/flat/c22286ad6268347c6...
Seatback entertainment in aircraft are not critical. Aircraft are.
In Postgres, there are plenty low-priority, low-impact bugs. #16265, for instance, reports that the capitalization of a relation in an error message is confusing. The report got mostly ignored. And that seems not a great loss.
Dare I say, solving every bug does not make economic sense.
Until it is. Swiss Air Flight 111 went down with 229 lives lost because of a new infotainment system whose electrical cables started sparkling.
Yes, not every part of a critical system is critical. Determining which is which is a complicated task, though, as parts interact. Not too long ago (and by this I mean within my lifetime), we were ignorant enough about our own bodies to think that our microbiota was almost irrelevant to our health and that bad teeth were mostly a cosmetic problem.
Back to IT, Heartbleed was caused by addition of a feature that looked as harmless as you could think of. But it wasn't.
Judging real critical-ness of bugs is hard. Yeah, capitalization of an error message is likely harmless, but I would treat, say, any buffer overflow as serious.
why would the optimal amount of bugs be anything less for a database?
In which case maybe leaving them alone is practical?
I love fixing bugs, but your post triggered this thought so I thought I'd share.
Remove pgrust: the entry is cheating (per-query kernels and result caching) (197 points, 10 days ago, 166 comments) https://news.ycombinator.com/item?id=50205816
For the other two claims in the removal PR, we wrote scripts you can use to verify the behavior of the original 0.4-preview binary:
- Per-query kernels did not serve the benchmark queries[0]
- The preview's fast repeat LIKE queries came from a dictionary filter cache, not a cached query result[1]
Across all queries and instance sizes on ClickBench, 0.4-beta is ~7% faster overall than the 0.4-preview binary that was removed. If 0.4-preview's numbers had come from hardcoded kernels or faked results, the open-source general engine wouldn't be beating it
[0] https://pgrust.com/downloads/archive/2026-09-clickbench-snap...
[1] https://pgrust.com/downloads/archive/2026-09-clickbench-snap...
This seems…good?
That it might disrupt the project's "traditional" release cadence a bit doesn't seem terribly bad in light of the alternative, which is shipping another major update with bugs which have apparently haunted the codebase for some time.
"We found a bunch of ways to make our product safer and better. We need more time to ship a higher quality product."
PICKET SIGNS
So unless it gets outlawed, I don't see people opting out of it
The article also concludes that overall, yes, finding more bugs is a good thing. Basically saying it’s the LLMs “fault” that the quality is improving, alongside a slight delay to the release.
No, they were not. The problems they faced were absolutely real, but the solutions they chose to attempt were stupid and counterproductive. Which is why they lost so hard, and it was not even a contest.
Socialism movement that refused to use machines and factories would inevitably lose to capitalists that embraced them.
Socialism movement that refuses to use AI will inevitably lose to capitalists that embrace it.
> In the end, I don’t think that the delay in shipping PostgreSQL 19 is particularly concerning. It is not a long delay in the grand scheme of things, and it emerges directly from the project’s concern for quality.
We will hopefully get over this wave and be able to move forward with much more secure software, but it's crazy right now.
LLMs finding unpatched vulnerabilities faster than they can be reasonably patched, bringing all other software development to a halt.
And even if the impossible occurs and LLMs hack everything that's connected to the Internet, so what? That would be inconvenient but it wouldn't cause doom. People would muddle through and eventually fix the critical problems, just like they always do.
I wouldn't be shocked if in 1-2 years non-security vetted software released into the wild will be hacked on arrival, like Windows XP without service packs used to be.
If you consider it to be productive work, it's a massive productivity boon happening right now. Of course, product managers might disagree...
Also, in the article they mention a significant number of patches for older versions as well.
It's not that the LLM code is that great, but that many overestimated the human code and the human review.
Also: most tests are positive tests. Tests rarely observe the negative space: that which is there but shouldn't be there.
looks at particledb
What they are really good at is writing boilerplate and reading API documentation / example code. Avoiding subtle errors is not what they're good at in my experience.
I think comparing the resources they spent at it would be a more apt comparison.
AKA if you tasked 150 mathematics departments to only focus about that problem, and stop doing things like teaching and researching anything else, you might get close to the amount of resources that these companies put into solving the problem to get a press release.
Read about the entire huggingface attack and why it happened.
This is one of the least talked about and most useful features in any harness I would say, the review feature. I have never not found issues (either security or implementation) running it, no matter the code base or language.
Some people will much appreciate the holes in Postgres that are being plugged. Many more will be bored with patch releases until new features rollout.
A software affecianado appreciates patches, and might do well to teach laypeople the benefits of a good defense.
Imagine that the whole city has to install iron doors and window grilles, because wooden doors and glass windows are not safe enough anymore.
That's a lot of work, but is it productive in the end? Probably not.
With AI it's the same. World was fine with lax security approach before. Today it is not fine. Tomorrow we might need to prove every code line before shipping, will that be more productive? Hell, no.
I don't think that's true.
In medicine, science often doesn't know how specific drug really is working. They just do a lot of tests and if these tests are positive enough, they ship it. I think it's good analogy. Biology is very complicated and it's out of our grasp to truly understand how various parts of it interact with each other on so many levels. But if something helps, it's stupid not to use it just because we don't fully understand it. Sometimes it bites. Sometimes it bites 30 years later. But we still follow that approach.
Software is similar. The complexity of the systems we build is enormous, when we're taking in account all the moving parts. I just read that someone emits radiowaves toggling memory bit and ex-filtrating information. How can you account for that? Probably you can't. But we still need software for good things.
SWE's are not a serious bunch when what they do can affect the entire world.
Obviously this doesn't work on the internet since most hacking comes from overseas out of our control. Which makes me wonder if there is a possible future where we just block packets coming from countries that don't lock up the hackers.
Not obvious to me. I am generally a fan of Cyberspace manifesto and the free Internet, I think it's great people can reach outside their ingroup and talk to strangers and ostensible enemies.
But, that doesn't mean we should give up physical security, the guarantee that the connecting party comes from a certain physical location (state, company). I think allowing connections from anywhere by default is fundamentally flawed security architecture, for certain things.
You cannot make defense a binary. We live in an adversarial world. It is required to survive.
You have a perverse view of product ivity
The algorithm is: if the threat model is valid, there are vulnerabilities and you either have some software everybody knows about or a gang of hackers specifically targeting you, you should patch.
The part of the industry that profits from security theater will want you to patch no matter what, even if your business goes down in flames due to downtimes and new bugs introduced by patched components. They won't remind you to assess the threat model first.
Digital threat models are very topologically flat. One exploit made in one place can be instantly copied and used anywhere.
Also once hacked you can be used as a base of operations to attack others. There is no neutrality on the internet.
> That's a lot of work, but is it productive in the end? Probably not.
Your imaginary scenario defines the solution to be pointless. If that same city was subjected to weekly riots smashing windows & doors, and people regularly died, stopping "productive" work to install things that keep people alive and productive? That's productive.
The point it, this was not happening before. LLMs made cheap cyber attack possible. But it didn't make the implementation of fixes correspondingly cheap. So now everyone needs to spend extra money/effort to install grills to cover where LLMs fall short.
For postgres, this absolutely makes sense though as long as people are willing to shell out the profits from LLM use to support postgres development. That is one way this could workout for everyone..
Then they have a pikachu face when someone smashes their car window to steal something worth $50 they left on front seat because “who would do that?”.
What is even better thief will most likely pawn this for $10 but for him it is effortless $10.
One way to think about it is that production allows you to pay for stuff. Not necessarily directly but by way of trading. It is the output you use to buy things like security.
Consider an apple farm. What does the apple farm produce to pay its bills? Apples. They could also dedicate resources to pears, &c, and that would add to their production -- to the goods they produce, the goods with which they (indirectly) purchase tractors, fence posts, &c. They could also dedicate resources to security -- more fences, &c -- and this may even be prudent. Maybe thieves are getting taller. This doesn't really add to production, though; at best, it maintains production at the status quo ante (assuming the fence is effective).
Security bugs are just bugs, as Linus likes to say. And the bugs are often also crash or correctness bugs (use-after-free for example usually causes segfaults). For the most part this actually is a productivity boon in the sense of getting higher quality goods out the other side.
To use your apple farm example, it would be like an apple farm where, if the apple rots, it's sometimes just gross but also sometimes it explodes and kills you. Would it be productive to reduce the rot rate of the bomb-apples? I think so!
If that same city was subjected to weekly riots
That's productive.
The software is better in the end but not all of it will be productive.
https://en.wikipedia.org/wiki/Parable_of_the_broken_window
Basically, if you accept this line of reasoning, then someone going around breaking windows and someone else going around repairing them can be seen as good because it creates jobs. But that would be nonsensical and create poverty. That's why economists focus on wealth creation not job creation.
Fixing security bugs is only needed because people exploit them. It creates work but it doesn't create wealth.
Protecting yourself from bad actors also doesn't create wealth but it protects wealth you already have.
You know cybercrime and APT groups are not bored kids in their parents basement, there are groups that are running automated tooling to find vulnerable companies and if possible take all their wealth.
Just look at Binance of course, but if you are small company there are also people targeting those especially with AI assistance nowadays.
It does not. Scroll down in the given article, and you will find war as another example.
As GP said, it's a terminology issue. War spending, including defense spending, is not productive, because it does not produce wealth. It does not mean that protecting yourself from enemy countries or criminal is useless, or not worth it.
Nobody claims investing in security is a bad idea. It's just not productive, in the economic sense.
it is distinct from question whether it is productive to defend from evil people
evil people + protecting from them is obviously net negative!
but it is not changing that of evil people are present then reasonable level of protection from them is needed, and setting it up is productive
It becomes productive only once your business stops for days because somebody brought it to a halt by breaking into the servers and deleting them.
By the way, I think that even without safety concerns seat belts are productive because they anchor the driver more firmly, so we can brake harder and turn faster (a lot of roundabouts here) and get to our destination in a shorter time. Of course this is against general road safety.
(1) We live in nations of laws. The threat of state violence deters most thieves.
(2) You don't actually keep most of your valuables in your home. The stuff the thieves want is (mostly) actually stored somewhere else.
Security work absolutely is productive if it actually prevents theft. Securing your home is the wrong analogy, because it doesn't really prevent meaningful losses. Securing the bank? Different ballgame.
Human-written code isn’t inherently buggy, and LLM-written code isn’t inherently bug free. What is irresponsible is shipping code you don’t understand, and that will only increase with LLM-written code. A major reason why so much of the world’s software sucks is precisely that we think of it in terms of what’s “economical” and not what’s good or right. What we need is to slow the fuck down and stop treating every facet of life like a transaction. Software is filled to the brim with dude bros who “move fast and break things” and actively exploit people as their business model and somehow are called “the brightest minds of our generation”. Our priorities aren’t right.
I am not doing it anymore.
Agreed however getting humans to understand probability is tricky.
I had this discussion with my son recently about considering the severity of outcome multiplied by the probability of that outcome been the way to look at some things.
Taking options off of the client’s table is the core responsibility here.
In the end, it's all business decisions. If it's something I don't like as a customer then it's to be infantilized.
(If you disagree then please don't complain ever again about LLMs refusing to help you reverse engineer a piece of code or explain how a virus works.)
I’m not saying that’s right, I’m saying that’s how a lot of businesses think.
i would say the recent slew of vuln reports/breaches making the news rounds are doing work!
This is the silver lining i suppose!
What makes you think that AI trained on slop is suddenly going to produce quality code?
But any kind of mistake, any kind of ambiguity and you'll be in for a wild ride. It usually takes a while for the rot to set in. I've had to revert - more than once - to many commits prior to undo a bunch of work that had me completely puzzled as to what the reason that particular fork was taken to begin with. Usually I end up figuring it out but definitely not always. In some cases this kind of wild modification fed back on itself until the situation was beyond rescue. I'm a bit quicker with such reverts now than in the past, for the first year or so I would try to save what could be saved, now I just restore and try again.
But if you're not sharp when evaluating what gets done there is a good chance that you'll miss it and then it can pop up way later.
What I've noticed as correlation (but I'm not sure if that is the root cause) is that non-local effects and stack effects are particularly hard as is everything that requires a working model of the world.
edit: this comment seems to make people uncomfortable.
Harness, prompt etc etc are all peripheral IMO, though people like to think their super long prompt, custom harness and MakeNoMistakes.md skills are what contributes to their special skill with AIs, but really models are just improving.
You can just trust the AI to generate a perfect answer I suppose (and why are humans slowing it down at all if so?), but if you don't expect that yet you have to put in the work.
I'm not sure who is studying this yet, but there has to be a statistical relationship between three factors of
1. intelligence of the human: the design patterns and architecture is already modelled in the mind, steering the AI becomes easy to know what to say next
2. quality of the codebase: the LLM can pick up the adjacent patterns and write code that looks and works cleanly.
3. quality of the prompt: the prompt keeps the AI focused and ignores the trivial side effects; keep the AI wandering away from critical path work; sometimes the AI can't make the decision what call site is next
In summation, what if claude is just a mirror pointing back at us? It certainly would explain wildly divergent opinions of opus 5.5
To me the order of importance would be:
1. The model - if you use a terrible model you'll get bad output
2. The codebase - this definitely forms the model for what claude outputs to some extent.
3. The human operator, what they accept and reject etc.
4. Things like prompts, skills, harnesses etc. The quality of the prompt.
So I think our positions are really just a different ordering of what is important rather than fundamentally opposed.
Re Claude being a mirror, if it is it's a particularly distorting mirror which often changes things in subtle ways and is determined to make people feel smart and addicted to its output - it has been trained to do so after all. I really don't think a mirror is the best analogy, because it's too independent and seductive to be a mirror.
There's a good example of the sort of thing they produce without good guidance as a dead comment on this thread as a reply to yours. Encouraging, positive and superficially plausible, but without any actual meaning or intention. The equivalent of that in code is what we need to be wary of.
And on the other side, I have seen AI produce exceptionally efficient, productive and effective code - the opposite of slop - when it is treated like a junior developer who will do any idiot thing, except when it is mentored by an actually experienced, professional developer.
The source of slop in AI code is not AI, its not the training data - its in the actual workflow of the human involved in its production.
If you just turn it loose on a codebase and ship it when it passes the tests you're asking for getting burned.
What's your workflow, out of interest?
Good one, that seems about right by my read of things.
Review is more important now than ever. My workflow has 3 review stages .. after analysis/specs/design, just prior to starting development (intake), and in the quality stages of testing and so on.
Well, there's also an after-release review, but the sales/marketing guys do that as part of their followups.
Bottom line: it's a good thing. My software will be better after that. Even if it's currently not good for my ego (and my health). But I'm glad for all the tools that help me improve quality.
And sure, I can just ask LLMs to fix the bugs, which works well half of the time, and produces bad slop in the other half. This is where human expertise and judgement is still necessary to avoid letting our code bases drift into LLM spaghetti hell.
While I did fix most of the things because while they might not have been exploitable they were bugs waiting to happen, in all but one case I haven't been very impressed by any of the LARGE IMPORTANT SECURITY HOLES that were found.
But that doesn't matter for me. It shows me real bugs and less than 1% false positives. Some bugs are truly nasty and urgent, most are trivial and not exploitable at all, but they're bugs and so I'm fixing them.
Mythos isn't mythically superb for being the best LLM model, but only for not having these annoying guard rails that make all others randomly refuse to cooperate when you need them the most. Whoever invented guard rails - please stop! - they just slow the good guys down in the race.
> This has been stressing me out
> the bugs follow me in my sleep
Sounds like you need some vacation.
If you don't mind answering, how come? Is it specifically that you dislike how abusable your code was, or something else?
To err is human, I know that as much as anybody. Before the advent of LLMs, there were several (human) audits of my code for vulnerabilities, but they found little, which I was proud of. And now, Mythos had a quick look and found hundreds and hundreds of bugs (many of which were critical and even more were really emberassing).
Of course, I knew there were bugs that just weren't found because the auditors weren't looking closely enough, didn't have enough time/budget etc., but this unspecific feeling is different from having all the bugs accumulated over a quarter century slapped in your face at once.
If you get a good auditor they can find a ton, but most are box checkers there to make the audit company as much money as possible.
After my Mythos experience, I was wondering if there is really still a market for human code audits. I guess yes, but the work will be different, just like coding (and marketing and so on...) is different now. You still need to have an expert to wield the new powerful tools.
We want software that remains robust in the long term, so that even when people are able to inspect the source code, it's still solid, rather than hiding the source code and hoping people don't find holes everywhere.
You would think so but medicine has faced this paradox since the development of modern testing technology.
You would naively think more testing is always better. More sophisticated practitioners then talk about Type I vs Type II errors and the problem of false positives.
But even if you were able to come up with tests that had 100% accuracy, there are times when you don't want to run that test on someone. The key distinction is that people only ever die of one thing and having a tumor that doesn't kill them is often better than risking the surgery to remove the tumor.
Similarly, there are some bugs you don't want to find since the harm from the people it affects is smaller than the opportunity cost of fixing it. Ignorance of the bug is better than knowing about it and having to make an affirmative decision not to fix it.
Logically, a bug that prevents someone from doing something and a missing feature that prevents someone from doing something are equivalent, they shouldn't be thought of as two separate activities but should be evaluated via the same framework. The list of all defects in a piece of software is infinite so it becomes impossible to erase all defects. It's about prioritizing with finite resources.
> You can't predict the harm from a small bug when AI can chain complex exploits and when your software is used in mission critical applications as a data store.
Yes, any bug can theoretically be part of an exploit chain, just like any tumor detected can theoretically be the tumor that kills you. This is why the detection paradox exists. Once you become aware of the tumor, you face a duty to try and mitigate it. But we know from autopsies that people die with all sorts of tumors that didn't kill them.
Similarly, if we bothered to do autopsies of software products after sunsetting to discover a comprehensive list of CVEs, we'd discover a ton of bugs we were better off not knowing about because they ended up not making an impact but we would have diverted resources if we knew about them.
This is not an argument that we should remain totally ignorant of all bugs, but a refutation of the "common sense" belief that more knowledge is always better.
I guess it really depends on the severity of the defect. Would it allow some really significant exploit? -> Important to fix.
Does the defect only prohibit one very esoteric use-case which basically never comes up (and was thus never reported by a human user and only now that we have an infinitely patient machine checking every completely unlikely code path is it detected) -> more productive use of your time to develop a new feature which would be used by more people.
That very much depends on got you define “bug” and how you specify your product requirements. Also, you have to assume the specs are correct.
The key thing you forgot, which makes this approach non-sensible, is the omniscience, without which you have no clue whether your posture towards a new bug-finding tool is good or not, since you don't know whether it will mostly find those bugs that are not worth fixing and the time wasted on fixing them is detrimental.
Similarly, how is ignorance better than informed decision when ignorance also includes missing critical security bugs?
This isn't an argument against all automated bug detection, just that it needs to exceed a threshold beyond just "It has found at least one real bug not found via other means" which is the seductive threshold people want it to be.
Go on - define a bar here.
> just that it needs to exceed a threshold
Same. Be specific.
Firstly, the cost of fixing a bug once you know it exists is usually minimal compared to the cost of discovering that bug. This is the inverse of medicine, where the cost of testing (at least in terms of health outcomes for the patient) is typically minimal compared to the cost of treatment.
Secondly, if you have an asymptomatic illness, this does not typically make you more vulnerable (hence why it's asymptomatic). Like you say, you typically only die of one thing, so if you know the patient won't die of a particular diagnosis because something else is more likely to kill them, then it might not be worth making that diagnosis. But for software, you can be simultaneously vulnerable to lots of things at once, and any of these vulnerabilities can be the one to "kill" you. Some vulnerabilities may be easier to find and exploit (or do more damage) but in principle any active vulnerability can be exploited.
Thirdly, having multiple vulnerabilities makes software more vulnerable overall, because those vulnerabilities can potentially be chained together to create more serious exploits than each vulnerability would have represented on its own. To a certain extent this can be true in medicine as well, but generally it isn't.
Fourthly, we can't cure death (yet?) but we can cure all the vulnerabilities that we've found so far. There is no known upper bound on the lifespan of a piece of software, and we know that older code tends to have fewer bugs in that newer code, presumably because those bugs have been fixed over time. This is different from health issues, where a given condition may be chronic or terminal and therefore at a certain point there's nothing you can do to fix it.
Finally, ultimately in healthcare it's your body, and your decision was to what you do with it (assuming you can find a doctor willing to do what you want). If you have a condition and you want it cured, even if it is completely benign and will have no impact on your life whatsoever, then you can do that. (Again, assuming you have the resources and the condition is curable). But a software vulnerability does not affect you as the developer of the software, it instead affects your users/clients/customers. Therefore there are different dynamics in play. If I sell you a tool and I know it could be vulnerable, but I don't try and find out what those vulnerabilities are, then you're ultimately the one on the hook when all your data gets deleted or leaked.
Boy, I wish!
> Like you say, you typically only die of one thing, so if you know the patient won't die of a particular diagnosis because something else is more likely to kill them, then it might not be worth making that diagnosis.
You can't know someone didn't die of X until after they died, that's the entire point of this dilemma.
As to your second point, you don't know exactly what a person will die of, but you can make statically sound estimates. You generally won't test unlikely potential causes of death (either because the chance of a positive result is too small, or because you know there are other, more serious conditions). But once a vulnerability is known, it is infinitely reproducible, which means the chance of the vulnerability being "fatal" (assuming the conditions exist for it to be exploited) is essentially 1.
The other flaw in the analogy is that in medicine, the patient chooses to do the tests on themselves. But in software security, there is an adversarial attacker as well as the "doctor" trying to defend the software. And in general, if the defender can find a vulnerability, they must assume that an attacker could already have found the same vulnerability (because the attacker and defender have largely the same tools and information at their disposal, especially in open source contexts), in which case, as discussed previously, that vulnerability also needs to be assumed to be fatal.
But what applies to humans doesn't apply to software. First, to pursue the metaphor, codebase can definitely die/be compromised/crash because of multiple reasons at the same time; second, there is virtually no risk, out of man-hours, in fixing a bug.
It seems like this option is being taken away by LLMs for all major and widely used open-source projects.
For better and worse.
[0] Write skew observed under serializable isolation: https://www.postgresql.org/message-id/flat/CA+COZaBo+ZpKgMvx...
[1] Serializable anomaly: duplicate primary keys: https://www.postgresql.org/message-id/flat/CA+COZaBOiiRPmEfX...
[2] SERIALIZABLE can commit write skew after SSI summarization: https://www.postgresql.org/message-id/flat/CA+COZaCtK=UQbeQw...
[3] Fix temporal foreign key validation of pre-existing rows https://www.postgresql.org/message-id/flat/CA+COZaChq_ukm2o7...
[4] Lost lock during toasting allows FK violation https://www.postgresql.org/message-id/flat/CA+COZaC-2-GQabq+...
[5] Partition pruning can incorrectly exclude a RANGE default partition https://www.postgresql.org/message-id/flat/CA+COZaDXrfTaBjLE...
[6] timestamp vs. timestamptz comparisons inconsistently ordered under DST : https://www.postgresql.org/message-id/flat/c22286ad6268347c6...
Very curious about the details and why it’s unfixable. Is it just because timezones are such a mess it’s logically impossible to satisfy every constraint?
[0] https://www.postgresql.org/message-id/flat/c22286ad6268347c6...
There are probably real world cases where dropping the TZ/offset provides meaningful performance/storage savings, but it hasn't been the typical case in my current sector.
At least they are UTC and not EST or PST...
This seems rather extreme. The pathological case only seems to make a given timestamp ~1h bigger, so you could still use an index scan, just with a bigger input followed by a filter that removes the timestamps that shouldn't be in the query.
question is, how long do you think this persists before there's only a long tail of minor exploits left?
(And hope performance stays in the same ballpark as the old code.)
I suspect both old codebases and new ones that have a very intentional architecture are both dangerous to let loose.
The agent doing the fixing must have not only an incredible map of the project and how it is put together, but the ability to reach into annotated examples in the past and refer to them as it goes about a fix.
I’m seeing that frontier models will not follow intended patterns or know when to deviate if their primary charge is “fix this bug.”
It is perhaps a better idea to have agents’ primary job be “maintain this codebase” and one way that they do it is to fix bugs, and “oh by the way here is a bug.”
(Performance can be part of 'correct', and can be just as hard to actually get right in specification!)
That era is over, compilers are used for almost all software, eliminating entire categories of bugs.
(Operating systems, by the way, are the exception to the rule for assembly: the Linux kernel today has an x86/x64 assembly source component around the same size as MSDOS 5.0: ~250kloc of it. Windows is smaller, based on the NT4 kernel leak at around 70kloc. As a percent of total they're much less, of course, but they also do much more. And have correspondingly more bugs, too.)
Entire classes of 'safety' bugs that just don't exist in modern compilers.
E.g.: forgetting to "POP" something you "PUSH"-ed, especially when it was a conditional allocation for a temporary.
Similarly, type errors were much more common, especially signed/unsigned, or accidentally overwriting things in scenarios that just don't occur even with ordinary C code. Function args were easy to push "correctly enough" that it "worked", but not really.
I'm tempted to say "no, it isn't". But I don't have numbers to back that up
SQLite has gone into a frenzy of bug fixing this past year, changing long stable processes to ease things up: opening a specific report channel just for these kinds of bugs, adding features to Fossil to help manage things on their end (and that's just the visible stuff).
And with all that, the latest release has unfortunately been the buggiest I've experienced so far.
I skipped the .0 because of a crash, and only released .1 after giving up waiting for a .2 that never came, with a patch to fix a serious performance regression.
I spent a month supposedly shipping old bugs no one had a problem with, rather than shipping new bugs that instantly crashed things or caused measurable regressions.
That's on a software with stellar testing, and a good engineering team known to favour stability over the new shiny.
So the amount of new code added is limited by processes and product management. Add to that that newly added code can get a lot more checks on it now, and new code will be more secure too.
I also don't think your main premise (adding holes faster than we can find them) holds up, but that's another matter. TL;DR I think we're in a high intensity period right now, but this time next year it'll be quiet in comparison - possibly quieter than before AI code scanners were a thing.
BTW, the Postgres community is also quite protective of their culture, so I expect this welcoming nature of the community to persist for a while.
I missed when AI became accountable for Postgres releases, when did that happen?
https://news.ycombinator.com/newsguidelines.html
I think of you as a good HN contributor, but you've been posting a lot of unsubstantive comments lately, and breaking the site guidelines in other ways too. Can you please stop doing this, and return to using HN in the intended spirit? Your substantive, thoughtful points are welcome, but not comments like this current one, nor these (to pick a few examples):
https://news.ycombinator.com/item?id=50280377
Postgres release is delayed due to the quality bar getting raised. New features are getting more eyeballs, and therefore leading to more rollbacks. Pre-existing releases are getting bug reports and backported fixes at much higher rate. Bug fixing might be taking away time from feature development, but does harden the codebase. LLM usage seems to have been a net positive.
And half the software products are worse than that.
Any organisation that said "let's move fast and break things" is now busy finding out what's actually broken.
If no attacker reaches system X you are not as exposed as with your internet facing frontend.
But, and it's a big but.
Each layer must be made as security as possible or one flaw can lead to all the layers below it collapsing.
Not reaching system X is just security by obscurity. People have learned the hardware just because something isn't directly connected to the internet that doesn't mean attackers can't reach it. For example clever data poisoning attacks can reach deep into organizations and breach systems you'd never expect.
Truly mature software doesn't get used, because people see low activity on github and assume it's "dead" or see no new features in years and assume that it's stagnating.
And the average product is solving much easier problems (in terms of low-level programming, which is where the worst bugs tend to be) than a high-performance database.
The process of learning C++ (and to a much lesser extent, C) is almost exclusively one of learning about the footguns and how to avoid them. It's a life-long process, nobody ever finishes it.
Anecdote, I did a travel across USA and when I was looking for a hotel, if you go from apple maps, to website, to try to book, the "load browser" in the maps... could end in states where you could not go back because back is get out of where you are and go back to the start of the map... and was a terrible experience... then you turbo charge it with "implement game changer UI improvements for the new bar on UI design" goal of some guy with enough power to make damage... and you end with softslop (or is it sloprosoft?)
"It’s possible that some of this jump is the result of committers making more mistakes this year than they did last year, but I don’t think anyone who works on the project really believes that’s the main cause. Rather, it seems likely that PostgreSQL committers were about as conservative in 2026 as they were in preceding years, but a lot more of the mistakes they made were quickly found by LLMs."
and
"And I think that’s also why we had so many late reverts, and why the release itself is delayed. Two factors are at play here."
so?
On our own 1.5M loc system, we've found and fixed thousands of bugs, security holes and circular dependencies in the last year. Code quality has increased, test coverage increased. We've also recently set our compiler to treat warnings as errors, and to be as strict as possible. It is absolutely amazing to see the gaps being filled, it would've been an impossible task for human devs and IDE's would've made it very uncomfortable. We've also ironed out all the bugs and failure modes in our infrastructure automation / monitoring. Our next project is to eliminate all 3rd party dependencies, and after that to rebuild our frontends completely using zero frameworks, not touching npm ever again after that.
So when I read this about postgres, I was very happy that they found and pursued this side quest. They can make the codebase bullet proof.
Genuine question, because I don’t know: Wouldn’t it suffice to vendor your current packages and “only” stick to maintenance from there on out, instead of rewriting it all?
This should still achieve the goal of not needing to touch npm, but presumably with less effort/tokens, right? Or is it just that you would expect the result of that process to be less clear and maintainable than a full rewrite?
Maybe you don't need all of Postgres - so provide 2,3,4,5,n versions. As long as every version is covered by (possibly shared) tests, then you can release what works, ignore the rest (bugs and all).
How's that going? Any lessons learnt you'd be willing to share?
Mmm what else. I prefer postgres so I'm very excited about this. So many foot-guns will no longer haunt us like on MSSQL and less db/table locks to try solve (pessimistic vs optimistic locks). Oh and we are going to save a gross amount of money per year, as we will also be able to switch off a couple of Windows Server VM's in the process. These puppies are costing us a ton (I think about half of our infra bill are licenses...).
Just create a branch, spin up a test db and try to port a small feature. Its very easy in most cases. Debian + dotnet + postgres = dream team.
Years ago I had to work on a server that had some sort of water utility software on it and it was a rather jank piece of software.
Before we started this journey (about a year ago), we spent a ton of time cleaning up our codebase, splitting large files (10K+ lines) into files that are no more than 2K lines, things like that. Then we let claude scan our important files and summarize them, add comments for future claude sessions. So we've let it built up tribal knowledge per file and the reasons for things to exist. We tell it all about our operating environment, the country and legal framework etc, so every little piece helps it stay honest and on track.
So to start, do something simple like: Scan the files in this folder for database code and check for queries that loads the entire result set into memory and/or queries that loads all columns. Suggest which queries we can filter better or reduce columns being transferred. Once done, also check real db row counts, compare them with the queries in these files and check if they might benefit from an index or partitioning/sharding. Also check for queries that loads data but their results are not used. Don't make changes yet, just do a deep investigation and report back. Feel free to spin up muliple agents to batch the scans.
edit: it goes without saying you need a language with strict typing and a good compiler (we use dotnet). Doing this on python/javascript would be a nightmare.
I still have yet to see this be efficient. The time I spent doing this is way less efficient than simply giving the LLM some high level direction and putting that time towards reading the code it generates and correcting small piece by piece (occasionally doing things manually, but that’s only for more off the rails stuff).
It’s about the same amount of time as up front planning, except I know the code much better as well as end up with less mistakes compared to having it implement a front loaded plan.
Knowing the codebase makes everything else 10x more efficient too - triaging bug reports, planning other work, knowing what work needs done (this is a big one), reviewing PRs, etc.
Let's hope this is the future of Software Development. There is so much to create yet.
How long, in your estimates, do you think you have before your career ends?
Companies will do different things, just like they did with mobile, the internet, and computers in general.
Some strategies will work, some will fail.
This is fairly common SV strategy, used by Tesla and others, and it works extremely well in spite of people pointing out the obvious weak points.
AI is getting so good, it will probably be both.
Would you mind elaborating?
Oh it also helps to split refactoring work into three buckets: business logic changes, performance changes (same output as before but faster) and cosmetic changes (rename files/folders/classes, breaking large files up, add logging/comments, domain layering etc). We tend to focus on performance and cosmetic changes for bulk refactors and leave logic changes to more dedicated sessions as to avoid changing what the system do (our domain is fairly complex and we cannot afford breakages). We also managed to port ALL of our code to dotnet 10 lts and have clean boundaries between projects (we have a mono-repo with about 50 projects in it), so we can wholesale upgrade dependencies without a sweat.
Of course, crashes and bugs can happen when applying quickfixes, which isn't nice. Ideally, they should be reported and fixed. You can also ask an agent to fix the occurences instead of using the IDE quickfixes if that's blocking you.
I guess what I'm trying to say is that IDEs shouldn't make that process harder, they can actually improve agents' results. For example, you don't need to write custom skills to prevent agents from writing some class of bugs if the IDE can detect it and agents have access to the results.
> Changing 200K+ lines of code yourself is a nightmare (and to prove it still does the same thing afterwards).
If an agent does it, you still have no proof, but okay.
Full disclosure, I work at JetBrains, but not in the IDE department.
Where now, the compiler flags & strict analyzer settings forces claude to do the right things (we explicitly tell claude not try and bypass or lower the barrier). I think it only works because we don't spend 95% of our time in the IDE anymore like a couple of years back, so we as developers no longer feel that friction, as claude tries to satisfy the compiler instead. Its a massive mental load that is just gone, we haven't had any conversations in meeting about this kinda stuff anymore (remember in the past we would have flameouts in meetings or code reviews about linter settings...).
Do you have a specific skill to guide an agent to scope the work of an agent so it stays in one particular bucket? Would you mind sharing it?
Overall, there's much less need for prompt skills anymore. They're still very useful for things like best practices, or things you need to make as reproducible as possible, but big skill and planning sessions aren't nearly as necessary anymore.
I could imagine scenarios where only well off developers and companies could afford that.
Then if you don't have the money, you are out of luck and risk having vulnerabilities in your code.
And it takes just one adversary to pay the bot & the bot will happily find an exploit for them in your code.
This worry this could lead to a serious power imbalance & furthe damage the free software community...
1. The cost of running AI is dropping down to "free" - there's open weight models, open source tools, and guides on running your own if the long list of suppliers isn't good enough for your use cases.
2. Vulnerabilities have always been a risk; AI is not a panacea. The pre-AI best practices still apply. Furthermore, a huge percentage of vulnerabilities are in 3rd party tools and systems, which are systematically being scanned and fixed and/or exploited, then fixed.
You mention "the free software community" in your last sentence though - this implies the code in question is open source, so it's able to receive plenty of improvements from anyone with access to these tools.
Worse case, we enter the Dark Forest era for technology. Read up on this concept and apply it to the tech industry. I suspect open-source might die and everything becomes private, no more public networks, no more open access, as to assure survival. In a sense we are halfway there already as it seems the internet is starting to close down, people not posting their art and music as much anymore, not posting their knowledge for free anymore, as it might be used against them by AI (the other party in the dark forest..). Thus private authentic experiences will become more valuable.
What a bullshit.
It's different scenario, different trade offs.
You mitigate risk of library being sold to malicious 3rd party, you avoid risk of your supply chain being hacked, you avoid API/behaviour breaks, etc.
The trade off is that you have to maintain it.
That's why dotnet's ecosystem is so loved due to MS providing 90% of needed libraries, so you avoid some 3rd parties maintained by single person.
Often it can be a very good choice to write the functionality yourself and not rely on external libraries.
And since we now have coding assistants, maintaining in-house costs near nothing now. Nobody else can break us, only ourselves.
The only "third-party" packages that we want to ideally allow/support would be provided from LTS versions of our runtime (in this case, packages provided by the dotnet team and audited by microsoft). So we will always try to build around what the framework provides. Second choice is our own code, third choice is open-source packages, fouth choice is a separate application/appliance (lets say ZenDesk for ticketing). Over time, we will reduce this footprint to only have dotnet team code and our own.
Remember most CVE's can be hidden on servers that are not publicly routable and don't have a real exploitation paths. The bits that are public (lets say a linux edge server with a web server on it, some firewalls, a couple of open ports) - these should be hardened to the max and not allow full access to all the services running behind it. We don't even give our internal services public IP addresses to make near impossible to route traffic to them.
Our git repo's are also not on public infrastructure (serious, think carefully about hosting on public services and if you really need it), never mind making the repo public.
We also have a dedicated secrets/certs manager, so no credentials are in codebases or pipelines. Very strict in this regard.
also some 3rd party libraries are overdesigned/overcomplicated because everybody wants the library to do everything for everybody.
This overcomplication can easily introduce security problems that is not present in the 1K solution.
Also a source of significant dependency churn even excluding security bugs.
Whereas with my own code I don't have to worry about scheduling work that's only arisen because somebody outside my organisation has decided to make a breaking change in a dependency. Paying that tax got old over a decade ago, and my patience with it now is close to zero.
The point being, if you weren't committed to maintaining your own implementation of a library before the advent of LLMs, having access to LLMs does not fundamentally change the calculus. The ability to generate code more quickly does not automatically grant an organization the capacity to maintain more code.
Of course, sometimes simpler is “I stored the passwords unhashed”, so you have to find the middle ground.
The thinking of "it's inside so we're safe(r) from attacks" is a false sense of security, as (e.g.) Maersk learned with NotPetya:
* https://www.wired.com/story/notpetya-cyberattack-ukraine-rus...
From the year 2000:
NAT (particularly NAPT) actually has the potential to lower overall
security because it creates the illusion of a security barrier, but
does so without the managed intent of a firewall. Appropriate
security mechanisms are implemented in the end host, without reliance
on assumptions about routing hacks, firewall filters, or missing NAT
translations, which may change over time to enable a service to a
neighboring host. In general, defined security barriers assume that
any threats are external, leading to practices that make internal
breaches much easier.
Kind of works until the TV, the fridge, a user's BYOD or your printer has infiltrated your LAN.
So it kind of doesn't.
To be fair, there is more nuance to your statement.
A lot of it depends on your programming language.
For example with Go, with its extensive and excellent stdlib, you would be foolish to rely on large swathes of 3rd party libraries, because so much can and should be done using the Go stdlib.
On the opposite end of the spectrum, you have Rust with an effectively non-existent stdlib.
And the problem you see frequently with Rust is projects importing half the internet, with a Cargo.toml the length of your arm with hundreds of dependencies. Of course the project has never reviewed its dependencies source code, and the project probably barely keeps its dependencies up to date. Its a supply-chain attack waiting to happen, not if but when.
So with languages like Rust, you have to take an approach of "minimal reasonable dependencies", make an active effort to keep your dependency list short and in-control.[1][2]
> less well audited than public libraries
I also struggle with this particular argument because the reality in open source is "well, its open-source, it has many eyes" is in fact mostly a dream. Since we all know 99% of the time its a small group of maintainers (assuming there is are even multiple maintainers[3]) and not many (if any) external reviewers.
[1] https://tweedegolf.nl/en/blog/119/sudo-rs-depencencies-when-... [2] https://tweedegolf.nl/en/blog/104/dealing-with-dependencies-... [3] https://xkcd.com/2347/
Security through obscurity protects you from low effort vulnerability scanners.
It makes you more vulnerable to a sophisticated targeted attack. A competent hacker or AI will quickly find vulnerabilities in your custom libraries.
I don't know were it will leave us in the long term and I don't think I'll like the answer.
Only where the AI solution is verifiable by the dev doing the promoting and review. Cryptography is probably better served by libraries vetted by experts.
That would be like investing in things one doesn't understand for themselves.
We routinely place trust in things we cannot personally verify.
Ultimately the guarantor of cryptography code won't be humans, it will be councils and red teams comprised of other LLMs.
So why wouldn’t a competent AI find those vulnerabilities during development? I don’t think the proposal is to have human devs hand write the custom libraries and yolo into prod.
It probably would. Except it’s internal, and not a reportable CVE, so the fix will be deferred.
(Defence in depth does mitigate this somewhat. The attacker has to be right two or three times)
Sure I shouldn’t roll my own crypto but I can write a logger…
Also any part of the program can send data to the logger. This means the logger can receive malicious payloads from anywhere. Maybe your logger handles ascii control characters badly?
Technically you can configure it to send to other places - presumably. In practice you'd be better just redirecting stdout, tailing the file, or installing a mod?
The only even slightly tricky part is like "don't mess unicode up", rest is straightforward. Sure it can be complicated with logging sinks, configurability, when to flush, monitoring and so on but in many cases you just don't need all that. You just build it whenever you do need it.
Capture log messages and write them directly to stderr? That’s easy.
This is where the LLM coded library replacements make sense. When you have non-critical code that can be easily swapped, I’m happy to let an LLM generate a logging package. If I have a need to send log messages out to a separate server for archival? I’d be a little more critical.
But crypto really is just a small thing. I’m not going to design a nuclear reactor but I’ll definitely do stuff that depends on the electricity from one. Similarly one can still do their own authentication; its not like even Apple forgot to rate limit…
> just in your own code that is less well audited than public libraries
why do you think public libraries, which change all the time, get deprecated, handed off and supply-chain attacked are well-audited in any way?
For example, a few years ago I removed a rather large 3rd party mathematics library because we were only using it for a single mathematical function that was very easy to replicate.
Remember, 3rd party libraries need to:
1: Be periodically audited to make sure you are in compliance with the license terms, and preferences of your company.
(1.5: This becomes important during an acquisition as someone will always come around doing due diligence.)
2: Be periodically updated (in case of security bugs)
2.5: Be updated on short notice when a security issue is revealed.
In the case of the mathematics library, replacing it with roughly 20 lines of code and 100 lines of tests is a significantly better tradeoff than constantly reviewing it every time one of the above things happens.
Now, at what point is it "worth it" to use the 3rd party library? In our case we'd need to make much heavier use of the library, probably at least 20 different mathematical functions from it.
If you spend a ton of time hardening your own library code and don't have much feature sure, then that can be a win.
It's also crucial to scan for "analogues" because otherwise many LLMs autoreport as "not vulnerable" since "this applies to Postgres/whatever and this repo isn't using it"
If you get hit by cyber guardrails you should use a local model to prune any CVE specific language from the reports.
To give examples, every single LLM (and people too tbh) will write netcode that's vulnerable to these CVEs: CVE-2026-27135, CVE-2024-45311, CVE-2021-32567, CVE-2019-9511; without explicitly mentioning the CVE text no LLM finds these either.
>set our compiler to treat warnings as errors, and to be as strict as possible
>ironed out all the bugs and failure modes in our infrastructure automation / monitoring.
>eliminate all 3rd party dependencies
>rebuild our frontends completely using zero frameworks
Oh yeah, that's what i call "software engineering" and excellent use of LLM's! Great to see Company's with a Mindset like that.
What's the proportion of bugs you've encountered that was due to a library versus due to your own code? I suspect it's probably more than 1:1000.
This is a stupid decision.
For me it’s less about bugs and more about supply chain security risk. And with AI and only building what I need from a dep, exactly how I need it, the cost of doing this is very low.
It's a nice idea, and in theory sounds great, but you may be trading 1 library with 1000 bugs for 1 million implementations each with one special bug...
You might say that's a big "if", but you did say "1 special bug"...
AI-slop reports are filling up the queue and taking away time from volunteers. So contributing to delay in releases/features.
> Code quality has increased, test coverage increased.
And that's true for many old codebases!
> We've also recently set our compiler to treat warnings as errors, and to be as strict as possible.
Yeah. Everybody who said it was impossible, that no budget would ever be allowed to fix old bugs, that new features "to support the business" was the only thing that mattered and other nonsense like that have been proven wrong. Arguably they were kinda right but then LLMs happened and that's what proved them wrong.
I love to see what's happening: at very long last people are understanding that security is something to be taken seriously. They're understanding that "more code = more bugs".
> Our next project is to eliminate all 3rd party dependencies,
I love the spirit!
> and after that to rebuild our frontends completely using zero frameworks, not touching npm ever again after that.
I hope many more shall realize that now we can give the middle finger to a great many pathetically bad and totally unfixable (even with the help of LLMs) ecosystems.
Does management appreciate it? Do customer see any value in it?
AI can help companies provide more features or cut costs.
This extremely non-obvious and non-transparent to managers and completely incomprehensible for customers.
Good. That's how it should be: do not trust motivated engineers, trust a text generator that has no clue nor responsibility instead.
/s
No. Debt is a tool. Sometimes it’s good to pay it quickly, but sometimes it’s better to pay it as slowly as you can.
You never wanna intentionally repay tech debt as slowly as possible. You might be able to tolerate/ignore it for longer depending on the circumstance, but you always wanna get rid of it sooner rather than later.
You can go with "eliminate all 3rd party dependencies [...] frontends completely using zero frameworks" and be stuck with JS. Or use a typed language (TS, ReasonML/ReScript or even Elm) and an API client generator (available for all those languages) and get type safe API calling from a more strongly typed language.
As much as I prefer not to use NPM and even more prefer (even my LLMs) barely touch JS. A bit of safety is cheap these days, as you rightfully pointed out.
Perhaps your workflows and processes are so efficient that this actually makes sense. But I’m still not convinced AI assisted development is at that point yet; I think the classic Build vs. Buy arguments still hold.
How much is this going to cost to maintain? How many dev hours? How many tokens? Is your in house solution going to provide the business and customer enough perceivable value that it’s worth the opportunity cost? Are you going to apply the same rigor and security testing that the popular frameworks are getting from the largest research firms in the world? How do you know the LLMs aren’t going to apply similar patterns and make the same mistakes? Why can’t you just use that AI expertise to accelerate patching so your exposure time is limited? How do you know you’re going to outpace the major frameworks and tooling providers in 2 or 4 or 6 years when they catch up with AI assisted development, if they haven’t already?
And very personally speaking, I’m still holding my breath to see if this AI bubble is a real thing or not. Because if it ever pops, well, that’s not going to go well for companies that are embracing this stance.
But the gp said it is, and gave examples.
The third party dependencies are also using AI to improve their code.
Just remember the golden rule of crypto is to never fool yourself into thinking you can write your own crypto libraries.
Wow, the innovation! I am so glad we have this AI thing!
> Code quality has increased, test coverage increased
And of course - two meaningless metrics.
A pattern I'm seeing w/ companies and projects not used to continuous delivery (or at least shorter release cycles) is that they are slipping in schedules more now.
I think in the current environment it makes a lot of sense to re-think how often releases get cut and how to handle security related updates.
When you do only one or a few releases per year everything is 'higher' stakes and more bureaucratic.
LLMs are a gimmick, btw. This is an Anthropic marketing stunt, how much did they pay you? /s
I mean, maybe just innovate less? "Innovation" is one of those words that makes me queasy nowadays. It's been true for decades that a good deal of what's touted as "innovation" is just stuff that is not actually as important as fixing what's already there. Most software releases new versions far too often for far too little reason. So overall it's good to see Postgres slowing down rather than just barreling ahead with a new release.
Step 5: Do you fix bugs before writing new code?
(now i feel old)
The fact that you found it with your LLM proves that it's not protected by obscurity anymore, either from you or from the other person who might want to exploit it. So it has to be judged by its severity alone, assuming that it's easily discoverable now because it probably is.
In binary any security flaw you've written is a risk the moment it's put to use. So, yea it's a real risk.
In probability measures it's much more difficult to figure out. If your application is something big that everyone uses and has high stakes then the probability is always high even for things that would be pretty hard to trigger. For example who knows how many nation states were sitting on postgres bugs for years? For smaller more private software AI has increased the risk greatly.
Either way it's irrelevant in my eyes. There was always going to be a day of reckoning in the future. Some new technology was always destiny to show up and rain on our lack of security parade.
This is a great way to see it
No, Postgres has bugs and needs to be propped up by AI because of incompetence. AI finds nothing in hundreds of projects.
So I'm not particularly surprised to see postgres experiencing the same thing.
And if uncovering those bugs means delaying a release to fix issues that might otherwise have gone unnoticed for years, it seems like a pretty good tradeoff.
But.
Something's lost in this need for perfection, knowing that every bug will be found quickly and cheaply if not fixed. It tilts the time spent towards security hardening rather than feature development — which I still agree is a _net_ good.
An analogy I find keen is going from living in a safe neighborhood where nothing happens if you forget to lock your back patio before bed, to one where if you ever once forget to lock your back patio, it will be instantly noticed — either by something benevolent, or something malicious¹.
I know this is indisputably "better." But something's lost and something's gained in living every day², and in this case what's lost the ability to be innocent and naive on the Internet with a low likelihood of repercussions. To be clear, there's always a baseline need to be careful, don't leave port 443 open etc, but you could run a simple webserver you wrote up as a toy project without too much concern.
Will the trend towards every gap being found nearly immediately result in better software, a safer world, and overall happier people? I don't think so.
-- ¹I saw this analogy somewhere on Hacker News recently, so I don't want to steal credit. I'll update if I come across it.
²Joni Mitchell, "Both Sides Now" (1966). Highly recommend a listen.
This is pretty lazy. Especially considering the fact that later in the post author talks about how humans have to author the fixes, because the AI ones aren't up to snuff. It's not a stretch to think that more of the AI generated stuff that's not up to snuff just got through. And, focusing on bug fixes is going to yield more but fixes, regardless of the technologies involved.
I'm not saying that LLMs aren't helping, necessarily, but this isn't very convincing.