Hacker News: davenz

New comment by davenz in "Lines of code got a better publicist"

davenz — Fri, 12 Jun 2026 06:14:28 +0000

Hey, author here. My AI overlord advised me not to engage with this comment, but I don't like being told what to do.

I don't know what the reference to "orange AI fanboy site" is, sorry, but I do find it interesting that what I assume is an AI-writing-detection-site pegged it as 100% AI written. Obviously folks will believe what they believe, but just to satisfy myself I will state that I used AI (Claude Fable 5) to help with research by suggesting and then summarising articles, then interviewing me both as a supporter and detractor to really solidify my position (essentially arguing/debating with the AI), and then it suggested the post layout (funnily enough including that specific subheading "Why I actually care", which I will concede sounds AI-as-hell), but I then wrote everything else. I did paste in any quotes and specific numbers (which that section has quite a few of), and I did then do a final pass with AI (both Fable, and then ChatGPT) to pick up spelling, grammatical issues, formatting, etc. Anyway, have a good weekend and thanks for reading. :)

New comment by davenz in "Lines of code got a better publicist"

davenz — Fri, 12 Jun 2026 06:03:36 +0000

Author here. Thanks, this is a useful addition I'd missed. The 40-point overestimate from their early-2025 work is exactly why I read METR's current position as "we can no longer cleanly measure this" rather than "AI definitely speeds you up now"; self-reports are doing a lot of load-bearing work in every direction, including in my own sense of my productivity.

New comment by davenz in "Lines of code got a better publicist"

davenz — Fri, 12 Jun 2026 06:00:37 +0000

Author here. I think this was my favourite (and favorite) thread. I chuckled. I'm also pretty sure I could find ways to weave "septic" into a post about AI.

New comment by davenz in "Lines of code got a better publicist"

davenz — Fri, 12 Jun 2026 05:58:21 +0000

Hi, author here. You're absolutely right! (yes, Claude-pun, I'll see myself out...) - I mean, I agree that Company A will remain "good", and may not need to worry about Company B. If that came across in the post then it was unintentional.

A lot of the criticism I've seen here could probably be addressed by me clarifying that no, I don't think AI should be blindly adopted, and no I don't think it adds efficiency or productivity by just existing and being "used". It needs thoughtful implementation, and each developer I know or manage or have chatted to uses it subtly differently, hence why I encouraged experimentation and trial-and-error, because when it "clicks" for you, then in essentially every case I've experienced or seen first hand, its added something (whether rigor, speed, efficiency, capability, etc).

New comment by davenz in "Lines of code got a better publicist"

davenz — Fri, 12 Jun 2026 05:43:59 +0000

Hey, author here. Fair criticism (and thanks for reading!). What I was reaching for with the "few months" claim wasn't about company-survival, rather individual practice. I didn't make that clear enough (I actually did write this, not "AI-slop it" as some elsewhere have suggested, so maybe I got "human-sloppy" by the end).

You're right that I don't back up the "woah woah woah" bit with any evidence, but I do stand by the sentiment that I think people should use AI; experiment, find the ways it can help you (I've found that there's a huge spectrum even among similar engineers, in terms of how they get value from it). Trying the tools properly costs you almost nothing, and "adopt deliberately, measure with the boring battle-tested stuff" is not the same position as "adopt or die".