Hacker News: knollimar

New comment by knollimar in "Windows 11 New Media Player Uses 3.5x More RAM, Charges for Popular Video Codecs"

knollimar — Sat, 20 Jun 2026 23:00:38 +0000

Whats wrong with 100MB*100?

New comment by knollimar in "GPT-5.5 hallucinates 3x more than MIT-licensed GLM-5.2"

knollimar — Sat, 20 Jun 2026 20:21:13 +0000

I make a trello board to direct spare tokens at when I'm bored now!

New comment by knollimar in "Iran requires insurance on ships using Strait of Hormuz, fees likely to follow"

knollimar — Sat, 20 Jun 2026 00:21:05 +0000

Americans are fueled by spite, though.

New comment by knollimar in "The room the economy can't see"

knollimar — Fri, 19 Jun 2026 13:45:47 +0000

I think there's no zero friendship making power, but matchmaking is certainly less.

I have online friends from matchmaking, but it's certainly much harder. I'm not underestimating it, but my language might have been too strong.

Consider the average young person's discord group.

New comment by knollimar in "The room the economy can't see"

knollimar — Fri, 19 Jun 2026 13:36:41 +0000

Even the types of games I believe are shifting to less third spaces. Matchmaking is replacing the MMO beginner town where a lot of the socialization happens.

I'd wager a lot of it is "to protect the kids" (read: remove socialization since platforms don't want to expend the effort to police bsd actors)

New comment by knollimar in "If your product is Great, it doesn't need to be Good (2010)"

knollimar — Fri, 19 Jun 2026 12:05:48 +0000

Oh man I want to vent about the Nintendo switch.

"actually shutdown long press + UI" process"

This plus if you plug it in after to charge it it turns back on

New comment by knollimar in "GLM-5.2 is the new leading open weights model on Artificial Analysis"

knollimar — Wed, 17 Jun 2026 22:38:18 +0000

I found sonnet preferable to k2.6 but 2.7 code for kimi seems better anecdotally

New comment by knollimar in "GLM-5.2 is the new leading open weights model on Artificial Analysis"

knollimar — Wed, 17 Jun 2026 17:01:52 +0000

Oic I misremembered OAI scores, I thought Sonnet had 51

New comment by knollimar in "GLM-5.2 is the new leading open weights model on Artificial Analysis"

knollimar — Wed, 17 Jun 2026 12:50:39 +0000

Isn't it closer to sonnet?

New comment by knollimar in "Humanity isn't ready for the coming intelligence explosion"

knollimar — Tue, 16 Jun 2026 06:14:21 +0000

We wait for gemini 3.5 pro and cope?

New comment by knollimar in "Show HN: Fata – Spaced repetition to fight skill rot from AI coding"

knollimar — Tue, 16 Jun 2026 04:18:56 +0000

Doing leetcode absolutely makes your system design better and you don't deploy algorithms you don't know.

Invert a binary tree is kind of a strawman for the absolutist claim that you don't improve the cleverness by practice, but it's an example that not all the practice is useful

New comment by knollimar in "Microsoft Teams introduces office attendance tracking via Wi-Fi"

knollimar — Tue, 16 Jun 2026 04:02:32 +0000

Do you need to enable location on your phone for this?

I don't want my device on work wifi at all.

I barely allow their email to hit it.

New comment by knollimar in "Curl will not accept vulnerability reports during July 2026"

knollimar — Mon, 15 Jun 2026 10:27:50 +0000

I've seen construction companies use all their vacatiom in December in America (since it sucks to work in in the cold)

New comment by knollimar in "Don't trust large context windows"

knollimar — Mon, 15 Jun 2026 02:56:14 +0000

Don't you resend after every turn, so splitting it avoids the n^2 token usage (granted it's cached so there's some optimal amount here)

New comment by knollimar in "Did Anthropic ask for this?"

knollimar — Mon, 15 Jun 2026 00:19:48 +0000

Yeah if the analogy is LLMs are doomsday devices it's hard to say mythos wasn't the best. That was my point

New comment by knollimar in "Did Anthropic ask for this?"

knollimar — Sun, 14 Jun 2026 23:38:48 +0000

>(They don't, but they claim they do.)

I hate to shill them, but wasn't mythos/Fable SOTA?

Your main point still stands without this aside

New comment by knollimar in "Not everyone is using AI for everything"

knollimar — Sun, 14 Jun 2026 22:12:04 +0000

instead of telling him out loud, "hey we see you're hedging and applying bullshit interview speak, your answer isn't sufficient", we asking in increasingly obvious but different ways.

It was an internal candidate so it would be awkward to tell him to his face he was floundering.

New comment by knollimar in "Not everyone is using AI for everything"

knollimar — Sun, 14 Jun 2026 22:05:17 +0000

It was a question of the form trying to figure out how you deal with "X", and he denied X having ever happened, despite that being a core part of his current role.

He was an internal candidate, we were interviewing him to see if we could trust him with more responsibility (more X specifically), since the new role shouldn't cover up X when it happens. The role involved doing X for himself AND for other people.

Similar to the form of "tell me about your biggest weakness" and you responding with "I have no weakness".

New comment by knollimar in "Not everyone is using AI for everything"

knollimar — Sun, 14 Jun 2026 20:06:41 +0000

I just interviewed a guy and all three interviewers asked him functionally the same question. He hedged 3 times and we just wanted an honest answer...

New comment by knollimar in "Not everyone is using AI for everything"

knollimar — Sun, 14 Jun 2026 19:14:15 +0000

What if the way you use AI isn't particularly important to you? Are you willing to sacrifice employment for a principle you wouldn't draw as a line in the sand?

Sometimes it's okay to say "I don't know" and it's okay to say "I don't care" and it's okay to say "It doesn't matter much to me".

Every interview is corpospeak where you infer the intended meaning of words anyway.