Maxim Krizhanovski

@darhazer.bsky.social

I write about software engineering, management and AI. I read much more than that. Creator of https://chess-upgraded.com

One under appreciated effect of the #Al coding assistants: it makes engineering work more resilient to interruptions. The gap between meetings is enough to craft instruction for the AI, and when you lose context due to interruption - you can read it back from what you wrote to the agent

If you have tried chess-upgraded.com and found the computer too weak / doing random, illogical moves… well the logs were saying that it fails to execute the search and fallbacks to a random move, but I wasn’t watching the logs 😅 Anyway, it’s fixed now

Chess Upgraded — chess variants built as real games

We build chess variants as real, polished, open-source games. Play Chess Upgraded: RPG edition (build pieces and level them up), Chess Upgraded (captures upgrade pieces), or Cannibal Chess (captures t...

chess-upgraded.com

With the raise of Al, it's important to remember than direction is more important than speed. If Al helps you find the right direction faster - great, use it. That's why prototyping with Al is so powerful.

I wonder if AI adoption will teach people anything about systems and constraints. It can give insane productivity boost to a single engineer, yet the larger the organisation, the smaller the overall gain.

People forget that sorting (as in SQL order by or Ruby sort_by) is not stable. Whenever you sort by a field that may contain the same value, and you need a stable order, you need a secondary sorting key that’s unique

One of the reasons I'm skeptical about the idea that AI "productivity" will lead to out-competing other companies is contained in this post. AI encourages adding adding adding adding (I just read a post excited about generating 150K lines of code in a weekend!!!), but "adding new stuff" != value

Mikal Lewis@mikal-el.com · last yr.

Product teams often assume the next feature will unlock success. But when every team depends on what’s not yet built, momentum stalls. I call this N+1itis. The strongest teams know: if you can’t win with one less feature, you can’t win with one more.