theptip

10 ms·

211.
by theptip 5mo ago
But also - anyone would be interested in purchasing orbital compute at the price Elon is quoting.
212.
by theptip 5mo ago
> They are waiting on the next well-formed spec Is this actually true? Maybe in a widget factory. I think it’s an anti-pattern for the new world. When you look at places that are shipping at insane pace (like Anthropic) the secret is not
213.
by theptip 5mo ago
I’m missing the premise. For internal apps why would you ever reach for Computer Use vs just having your agent whip up a cli or MCP? _of course_ computer use is worse. It is your last resort. Do not use it on state that lives in a DB that y
214.
by theptip 5mo ago
By you want to be able to interject “hold on…” and have it immediately stop talking, when it goes off the rails. And GP is correctly pointing out that the only negative here (silence waiting latency maybe being too low) is tunable separatel
215.
by theptip 6mo ago
Compared to most US airports, yeah the food and environment is excellent. I wouldn’t say the airport is worth a visit vs. SF itself of course.
216.
by theptip 6mo ago
No the point is you know exactly where T is just by looking at the dictionary (or at least, you learn this if you use a dictionary a lot). IOW your prior on the data distribution lets you skip the first 4-5 binary chops.
217.
by theptip 6mo ago
I enjoy speculating on what could come from revisiting the first principles with our new tech stack. The Fediverse makes it much easier to broadcast updates in a truly decentralized fashion. Maybe this can be the discovery layer instead of
218.
by theptip 6mo ago
TIL! I had assumed that veto was purely diplomatic muscle. That seems like a potentially very cunning soft takeover, in that case. Still, I think onshoring is strategically wise in a world where the US is actively antagonizing the EU.
219.
by theptip 6mo ago
It does seem like the USG should fund a few $10B to build out the alternative stack, given the strategic sensitivity. Best time to have started would have been 20y ago, second best is today. This is small change in the military budget.
220.
by theptip 6mo ago
Supply and demand - if you think it’s not worth the price, take your dollars elsewhere. This is the brutal reality; even with the crazy reliability issues, demand is still far outstripping supply at the current price.
221.
by theptip 6mo ago
This one seems clear cut as a HIPAA violation. Glad to hear that interpretation was upheld. However, regardless, we really need to just kill the data broker business model. Speaking as someone who implemented GDPR for my startup when the la
222.
by theptip 6mo ago
Yes, and - there is also something about the visceral feeling you get when your turn comes up in standup and you didn’t update any tasks and you don’t know the status of the thing you promised for this week. If the PM does the task list and
223.
by theptip 6mo ago
The Stasi would be the obvious cultural context. In the US of course the government buys this sort of information legally from corporations.
224.
by theptip 6mo ago
More context on what’s going on with LLMs solving Erdos problems: https://www.dwarkesh.com/p/terence-tao TLDR, most of what is getting solved so far is “easy” problems that were not seriously looked at by experts, and
225.
by theptip 6mo ago
The Eval problem; “Alice is supposedly smarter than Bob, but they can both tie their shoes just as fast”.
226.
by theptip 6mo ago
Can you share more on this market cap risk? I see legit stability / correlation risks but can’t work out the market cap risk mechanics. The cash was just sitting on their balance sheet not increasing Google’s valuation, turning it into
227.
by theptip 6mo ago
Rather than banning gambling I think you need to ban congress critters from trading. Polymarket is a quick and anonymize way of making long bets on your inside information. But there are plenty of other stock-based bets they already do make
228.
by theptip 6mo ago
Both are true. No sympathy for this guy if he’s guilty as charged. But also don’t forget that this guy’s trades are a drop in the ocean compared to the rest of the likely insider trading that’s visible in the Polymarket logs. (Eg on timing
229.
by theptip 6mo ago
A sacrificial offering to the public, don’t look too closely at the $1B of Polymarket insider trading coming from higher ups in the admin.
230.
by theptip 6mo ago
The tension here is that what customers need to reproduce is this result on their own problem. To measure this you need extensive evals on private data. OpenAI simply won’t share the data you need to reproduce this in the way you’d hope for
231.
by theptip 6mo ago
> if I needed a painkiller, I’d try to take ibuprofen rather than acetaminophen. My logic was that if eight grams of acetaminophen could kill my liver, then one gram was probably still hard on it. This article is explicitly considering u
232.
by theptip 6mo ago
It’s not bad though, it’s crazy good in comparison to any model older than 1y old. If you don’t have access to any vibe coding at all it’s gonna be life changing. But I think you are right, as long as Codex and Gemini are cheap alternatives
233.
by theptip 6mo ago
It’s pretty reasonable to say “demand is way up, quality is up, supply is constrained, and so price needs to rise”. It seems weird to segment this way though. Surely it’s better to just give Sonnet to your bottom tier, rather than cut out t
234.
by theptip 6mo ago
You can build evals. Look at Harbor or Inspect. It’s just more work than most are interested in doing right now.
235.
by theptip 6mo ago
This is the right way of thinking end-to-end. Tokenizer changes are one piece to understand for sure, but as you say, you need to evaluate $/task not $/token or #tokens/task alone.
236.
by theptip 6mo ago
I agree to disagree. However I do feel a lot of this comes down to facts about the world now, eg whether Claude Opus is doing anything interesting, which are in principle places where you could provide some evidence or ideas, along the line
237.
by theptip 6mo ago
> then everything people try to do with those last and most capable models would end up uninteresting I believe that some of my made up examples won’t end up getting built, but my point is that there is _so much_ low hanging fruit like t
238.
by theptip 6mo ago
This is thoroughly debunked at this point. The frontier labs are profitable on the tokens they serve. They are negative when you bake in the training costs for the next generation.
239.
by theptip 6mo ago
If all AI progress somehow immediately halted, the models that have currently been built will still have more economic impact than the Internet. Not least because the slower the frontier advances, the cheaper ASICs get on a relative basis,
240.
by theptip 6mo ago
I normally agree with this, but they objectively did lower the default effort level, and this caused people to get worse performance unexpectedly. And it does seem likely to me that there were intermittent bugs in adaptive reasoning, based

More ›