Hacker Newsnew | past | comments | ask | show | jobs | submit | simonjuk's commentslogin

I have this same frustration. The way I use Claude is for it to make edits one small chunk at a time so I can review the code as it writes, and to make sure it's going on the right direction. I find this much easier than generating a large pr I need to review.

More and more I feel like I'm fighting Claude: it edits multiple files using python scripts and it edits in parallel so if I stop it, I have to stop each individual edit. Both of which I've told it not to do repeatedly, but it still does then apologies which is worse.


This is solvable. Opus has adhd. Myself and others solved this by building ticketing and task management systems. The task constrains the agents actions and focuses the effort. Every update is catalogued. It’s easily reviewed and referenced. Opus doing terrible things and apologizing is not surprising. I’ve moved to a system of multiple agents reviewing and authoring changes. Convergence generally always results in a positive commit.

We recently recruited two developers. Our recruitment process involves asking three questions before the interview stage. Almost every single answer was written by AI - they all followed the same format and presented the answer in the same way, customised to the applicant.

It was mind-numbing. The non-ai applicants were obvious and really stood out as the best answers.


In my experience, there are two ways to use AI: speed or quality. Speed is where you give the AI a task to do and you review it; quality is where you write the code yourself and you get AI to review it. Both are valid for different situations.


My plan for bigger things is mostly:

Generate multiple solutions- they do not to work 100% correctly. And than I check which I would prefer. Which is more to our applications taste.

And than I would take the vibe output as a kind of a ‚plan‘ which I use to implement but not follow 100% and at the end I take my solution and review it. I gain speed with that because I often can quickly see the pros and cons of a solution way better than when I would manually do it and hang on a major roadblock and also I even see such roadblocks in the vibe output - it’s mostly the part with an unnecessary amount of new code that looks nonsensical.


Using a LLM whose output is slowed to the rate of a human programmer as a pair programming partner is a very interesting experience.


I work with public data, and I'd love to get access to this data, but I suspect that although you have scraped the data from public websites, there are licensing and copyright implications for actually using it.

See also the open addresses project by Data Adaptive [1] which is using Freedom of Information requests to publish public council tax address data. The problem they have run into there is that their address datasets are derived from proprietary Ordnance Survey data.

It looks like data.gov.uk is in the process of standardising the planning application process, and publishing them under OGL [2].

[1]: https://www.owenboswarva.com/blog/post-addr44.htm [2]: https://www.planning.data.gov.uk/dataset/planning-applicatio...


Thanks and yeah, some of my boundary data (for the choropleth) comes from ONS open boundary files which I think are OGL but I'd need to check the chain of derivation. On the data.gov.uk standardisation, I've seen it but last I looked it was policy and boundaries, not actual decisions. Has that changed? If they're publishing decisions under OGL I'd gladly ditch the scraping for a proper feed. On licensing more generally... I haven't fully nailed it down. Showing aggregates and pointing back to source, but yeah there's a gap between data is public and do whatever you want with it commercially.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: