tensegrist

oh i thought this was going to be a jpeg that loaded the highest-frequency data first and added the lower frequency data later

only insofar as it is 1. not illegal to do so or 2. the cost-benefit of violating such laws makes sense for the majority of users, who are not doing things that are actually illegal

because without such a critical mass of normal users you get something like tor or grapheneos that the state begins to associate with people engaging in unsavoury activity

part of Capgemini Invent

rip

things like that indeed tend to be fraught

some relevant links

https://arxiv.org/abs/2408.02442 https://arxiv.org/abs/2510.15061 https://arxiv.org/abs/2604.13006

that's like 6k tokens over 12 hours which doesn't seem like a lot for…any purpose i can think of

can't you use font shaping rules or whatever it is to essentially begin a span of text with a "private key" that then causes the rest of the message to render correctly by combining with it (a trivial version could be e.g. a rot-N based on a given N)

essentially, yes

it's not about the prose quality, but about the fact that it seems like the service of deleting ai-generated code is also going to be ai-generated, or at least the service is being advertised via ai-generated text, which is…funny

"created with", à la "built with"

Then we cut: the fourteen date formatters become one,

something's off here

the list numbering in that first link is a bit surprising

Here’s the part I need you to sit with

no, i don't think i will

i've been using nixos on an m2 air for a year now, the kernel is enough

there was a vibecoded prediction market–style page that was put up yesterday (?) that got the date exactly right i think

can a unique fingerprint (no pun intended) be extracted from hand geometry

at least 50% of the front page would disappear if this were enforced

on some queries the gaming card actually beat the H100 (1.26s vs 1.77s on Q10), despite the H100 lacking RT cores

despite?

aiui nvidia has buyback agreements covering these to prevent them from being sold off

i don't know why you're trying to analyze the meaningfulness of sentences that are not the results of a human thought process but are clearly rhetorical flourishes from an llm that "feels" compelled to fill its prose with them

On the Intelligence vs. Cost per Task Pareto Frontier: GLM-5.2 is on the Pareto frontier of the Intelligence vs Cost per Task chart, with the lowest cost per task among models at its intelligence level. GLM-5.2 costs ~$0.46 per task, compared to GLM-5.1 ($0.25), Kimi K2.6 ($0.31), MiniMax-M3 ($0.18) and DeepSeek V4 Pro (max, $0.05)

am i missing something?

When you invite guests in, you say "Please, come up!"

japanese also does this idiomatically: the verb 上がる "to rise" can be used for "to enter (a house)", which gives rise to expressions like お上がりください for "please come in"

blindsight is really something

for those not in the know

https://en.wikipedia.org/wiki/Commonwealth_of_Independent_St...

But it would take a writer of rare skill – which Nguyen is not – to make it compelling reading.

jesus

…bounty hunting?

It is very difficult to be a productive editor without custom tooling at this point

this is extremely reminiscent of the stackexchange situation

there was some interview where he illustrated the idea of a twin prime pair using 27 and 29

reasoning itself just affords the model a ton of extra forward passes / "time to think"

the, como se dice, "misalignment" between the content of reasoning tokens and the actual output following the end of the reasoning is a separate problem, extensively studied by e.g. Anthropic

reminded of Arthur Whitney's stuff

https://news.ycombinator.com/item?id=45800777

"Why it's wrong"

on the one hand this sucks on the other hand let a thousand schools of steinberger/strandberg-style weirdness bloom