convexly

5 ms·

31.
by convexly 6mo ago
That's a good point, I might have gated it too hard. I'll open up the full results now. Appreciate the feedback.
32.
by convexly 6mo ago
That is definitely a bug, thank you for pointing that out. Should have been neutral! I'll push a fix for this.
33.
by convexly 6mo ago
Just need practice! People have no idea how overconfident they actually are.
34.
by convexly 6mo ago
The Brier score and "diagnosis" are shown immediately, no signup needed. The email is optional and only if you want to see the calibration curve and the question breakdown sent to you. I'll make that clearer!
35.
by convexly 6mo ago
Update: 400+ quiz takers now... insane. Best Brier score so far is 0.007 (nearly perfect calibration). The worst came in at 0.600. Average is 0.230, still just better than a coin flip. Where did you land?
36.
by convexly 6mo ago
As much as it is counterintuitive, that is actually a valid calibration strategy. If you notice the questions lean slightly towards counterintuitive and adjust for it, that IS better calibration! But you raise a fair point about framing bia
37.
by convexly 6mo ago
Interesting data from the quiz so far: 160+ quiz takers! The average is 0.239 (barely better than a coin flip at 0.25), but almost everyone indicates they are confident in their answers.
38.
Show HN: I built a 2-min quiz that shows you how bad you are at estimating(convexly.app)
20 points by convexly 6mo ago|66 comments
39.
by convexly 6mo ago
This pricing only really makes sense if the users can predict their usage, if not people that use this heavily are just going to be hamstrung and are going to start rationing their usage.
40.
by convexly 6mo ago
The move to Linux is looking more and more appealing. These decisions are killing Microsoft by a thousand cuts.
41.
by convexly 6mo ago
At face value this seemed cool, but the more I think about it slides or prototypes are the same thing, just a different kind of theater.
42.
by convexly 6mo ago
That IPO valuation is starting to look even more shaky now that they're bundling AI (and forcing?) bookrunners to buy into a weaker business. Looking like a capital raise to prop up his other businesses.
43.
Ask HN: Are Brier scores useful for personal decision making?
1 points by convexly 6mo ago|0 comments
44.
by convexly 6mo ago
Agreed, pixel density at that size in 4k is a treat.
45.
by convexly 6mo ago
Went from ultrawide back to my 27 inch monitor and definitely feel more focused. Having everything open "just in case" was killing my output. Nothing alt+tab can't fix.
46.
by convexly 6mo ago
I like the approach of running everything locally. I'm strongly of the opinion that the privacy angle for local models is going to keep getting stronger and more relevant. The amount of articles that come out about accidents happening
47.
by convexly 6mo ago
There's something really satisfying about reading a 1997 paper and seeing that it is still completely relevant. The fundamentals haven't changed but the scale at which we can apply them has.
48.
by convexly 6mo ago
The flip side is that good ideas with honest framing often lose to bad ideas with better marketing. Being right isn't enough if you can't communicate it and most people don't have the patience to evaluate the honest version.
49.
by convexly 6mo ago
All the advanced engineering in the world and you still need to figure out how a toilet works in zero gravity.
50.
by convexly 6mo ago
Large orgs make decisions that prioritize short-term metrics over long-term quality all the time and nobody tracks whether those tradeoffs actually paid off. The decision to ship fast and fix later sounds reasonable in a meeting setting unt
51.
by convexly 6mo ago
The cadence point is understated. SpaceX launched 130+ times in 2025. The next closest was around 15. That's not a gap that closes in 2-3 years even with heavy subsidies, because it's not just the rocket, you need to account for t
52.
by convexly 6mo ago
The fact that someone had to build a separate app just to get the version of Instagram from 5 years ago says a lot.
53.
by convexly 6mo ago
Do you expose the posterior probabilities anywhere so you can see how confident it is in the result?
54.
by convexly 6mo ago
The question for me is just timeline. The "rest of the galaxy" revenue is decades out, while Starlink revenue is now. Most of the 1.75T has to be priced on Starlink working at telecom scale. If not, 1.75T seems like a steal for th
55.
by convexly 6mo ago
Did you try disputing through your payment method or is that not an option with how Swappa processes things?
56.
by convexly 6mo ago
The valuation only makes sense if you price in Starlink becoming a top 3 telecom and Starship opening up entirely new markets. Possible, sure, but the launch business alone doesn't get you anywhere near 1.75T. They're betting the
57.
by convexly 6mo ago
I mean it is a similar loop. Define what good looks like, measure how far off you are, iterate. I would say though that the people who've been doing that for years just have a head start that prompt engineers don't.
58.
by convexly 6mo ago
I'm genuinely waiting to see at what the valuation lands at. The gap between what SpaceX charges per launch and what everyone else charges is so wide that the moat basically is the rocket. Hard to compare against anything even now.
59.
by convexly 6mo ago
For most companies an $850B valuation means you figured out the business model. Here it seems to mean you convinced enough people that figuring it out is inevitable. Those are very different bets.
60.
by convexly 6mo ago
1000 points if your "proprietary model" is GPT-4 with a system prompt and you call it your "research lab"

More ›