5 ms·
The problem with this is none of this is production quality. You haven’t done edge case testing for user mistakes, a security audit, or even just maintainabili
by tannedNerd 9mo ago
The problem with this is none of this is production quality. You haven’t done edge case testing for user mistakes, a security audit, or even just maintainability.
Yes opus 4.5 seems great but most of the time it tries to vastly over complicate a solution. Its answer will be 10x harder to maintain and debug than the simpler solution a human would have created by thinking about the constraints of keeping code working.
- LatencyKills 9mo agoAgree... but that is exactly what MVPs are. Humans have been shipping MVPs while calling them production-ready for decades.
- maherbeg 9mo agoThat may be true now, but think about how far we've come in a year alone! This is really impressive, and even if the models don't improve, someone will build skills to attack these specific scenarios. Over time, I imagine even cloud providers, app stores etc can start doing automated security scanning for these types of failure modes, or give a more restricted version of the experience to ensure safety too.
- afavour 9mo agoThere's a fallacy in here that is often repeated. We've made it from 0 to 5, so we'll be at 10 any day now! But in reality there are any number of roadblocks that might mean progress halts at 7 for years, if not forever.
- christophilus 9mo agoEven if progress halts here at 5, I think the programming profession is forever changed. That’s not hyperbole. Claude Code— if it doesn’t improve at all— has changed how I approach my job. I don’t know that I like this new world, but I don’t think there’s any going back.
- usefulposter 9mo agoThis comment addresses none of the concerns raised. It writes off entire fields of research (accessibility, UX, application security) as Just train the models more bro. Accelerate.
- maherbeg 9mo agoBoth accessibility, and application security are easier to build rules + improved models for because they have pretty solid constraints and outcomes. UX on the other hand is definitely more challenging given how much of it isn't quite codified into simple rules. I didn't write off an entire field of research, but rather want to highlight that these aren't intractable problems for AI research, and that we can actually start codifying many of these things today using the skills framework to close up edges in the model training. It may not be 100% but it's not 0%.
- adriand 9mo ago> Its answer will be 10x harder to maintain and debug Maintain and debug by who? It's just going to be Opus 4.5 (and 4.6...and 5...etc.) that are maintaining and debugging it. And I don't think it minds, and I also think it will be quite good at it.
- jonas21 9mo agoI can generally get maintainable results simply by telling Claude "Please keep the code as simple as possible. I plan on extending this later so readability is critical."
- tannedNerd 9mo agoYeah some of it is probably related to me primarily using it for swift ui which doesn’t have years of stuff to scrape. But even with those and even telling that ios26 exists it will still at least once a session claim it doesn’t, so it’s not 100%
- bgirard 9mo agoIt's not from a few prompts, you're right. But if you layer on some follow-up prompts to add proper test suits, run some QA, etc... then the quality gets better. I predict in 2026 we're going to see agents get better at running their own QA, and also get better at not just disabling failing tests. We'll continue to see advancements that will improve quality.
- zamalek 9mo agoI think someone around here said: LLMs are good at increasing entropy, experienced developers become good at reducing it. Those follow up prompts sounded additive, which is exactly where the problem lies. Yes, you might have tests but, no, that doesn't mean that your code base is approachable.
- cyberpunk 9mo agoYou should try it with BEAM languages and the 'let it crash' style of programming. With pattern matching and process isolated per request you basically only need to code the happy path, and if garbage comes in you just let the process crash. Combined with the TDD plugin (bit of a hidden gem), you can absolutely write production level services this way.
- layer8 9mo agoCrashing is the good case. What people worry about is tacit data corruption, or other silently incorrect logic, in cases you didn’t explicitly test for.
- vbezhenar 9mo agoYou don't need BEAM languages. I'm using Java and I always write my code in "let it crash" style, to spend time on happy paths and avoid spending time on error handling. I think that's the only sane way to write code and it hurts me to see all the useless error handling code people write.
- kid64 9mo agoDepends on the audience
- aschobel 9mo agothere is are skills / subagents for that something like code-simplifier is surprisingly useful (as is /review) https://x.com/bcherny/status/2007179850139000872 https://x.com/bcherny/status/2007179850139000872
- structural 9mo agoYes, but my junior coworkers also don't reliably do edge case testing for user errors either unless specifically tasked to do so, likely with a checklist of specific kinds of user errors they need to check for. And it turns out the quality of output you get from both the humans and the models is highly correlated with the quality of the specification you write before you start coding. Letting a model run amok within the constraints of your spec is actually great for specification development! You get instant feedback of what you wrongly specified or underspecified. On top of this, you learn how to write specifications where critical information that needs to be used together isn't spread across thousands of pages - thinking about context windows when writing documentation is useful for both human and AI consumers.
- sksishbs 9mo agoThe best specification is code. English is a very poor approximation. I can’t get past that by the time I write up an adequate spec and review the agents code, I probably could have done it myself by hand. It’s not like typing was even remotely close to the slow part. AI, agents, etc are insanely useful for enhancing my knowledge and getting me there faster.
- ncruces 9mo agoHow will those juniors ever grow up to be seniors now?
- LinXitoW 9mo agoEven better. Job security for current seniors.
- ncruces 9mo agoThis makes no sense. Not even from a cynical and selfish view point. I consider my job to be actually useful. That I produce useful stuff to society at large. I definitely hope that I'm replaced with someone/thing better; whatever it is. That's progress. I surely don't hope for a futre where I retire and medics have access to worse tech than they have now.
- pseudosavant 9mo agoIsn't it though? I've worked with plenty of devs who shipped much lower quality code into production than I see Claude 4.5 or GPT 5.2 write. I find that SOTA models are more likely to: write tests, leave helpful comments, name variables in meaningful ways, check if the build succeeds, etc. Stuff that seems basic, but that I haven't always been able to count on in my teams' "production" code.
- joelthelion 9mo agoDepends on the application. In many cases it's good enough.
- mikert89 9mo agoIts so much easier to create production quality software