Nikhil R

went from being called a scam to Apollo and Blackrock buying ICO coins lmaooo.

The faster people can get to something 80% done and unstable, the more they are going to crave stuff that's already 100% done and reliable.

@status_effects @AlecRad @DavidDuvenaud This pretty cool work. Why 1931 in particular? Why not say 1960 before DNA structure discovery etc etc

New work with @AlecRad and @DavidDuvenaud: Have you ever dreamed of talking to someone from the past? Introducing talkie, a 13B model trained only on pre-1931 text. Vintage models should help us to understand how LMs generalize (e.g., can we teach talkie to code?). Thread:

@manasjsaloi Hey folks! Please DM me a small blurb and your LinkedIn.

Come join @manasjsaloi, me and our founding team to build the future of prediction markets DM me with the best AI tool or project that you have built along with LinkedIn.

@ai @gyani1595 ^

Well, this is something. Someone built a "SaaS Death Scanner" that rates companies on their probability of being replaced by a Claude Skill, which is just a .md file. It's a gag, but the underlying taxonomy (that it accidentally produces) is actually useful. The survivors all have something in common: physical atoms, regulatory capture, or infrastructure you literally cannot fork. The ones sweating are mostly software abstractions sitting on top of workflows that language models can now perform directly. Microsoft's comes with the best note: "You can't replace Microsoft with a .md file, but Microsoft is absolutely trying to replace YOU with one." @DeathByClawd deathbyclawd.com

@gyani1595 😂

@0xhelios @danrobinson Hey, We are building a prediction market and would love to have someone with your skillset on our team. We raised recently and are hiring. Interested to chat?

catching up @danrobinson

new from paradigm: we are building a tool for exploring prediction market data try it out today. I bet you'll find new markets you never knew existed

@TheBlock__ But it’s the bridged version. Not the actual minted one no?

Mizuho upgrades Circle shares outlook citing Polymarket’s use of USDC for settlement theblock.co/post/387431/mi…

@primexmpls I have 3k hrs on this game and I am closer to an absolute beginner than to a pro lol

Rocket League is crazy because anyone with over 2000 hours can instantly tell how this will go.

@manasjsaloi So you are saying the algorithm works fine?

From the wonderful newsletter by @eightyhi Add the internet also to this list!

@karpathy tptacek is great. I learnt a lot of crypto (not the coin kind) from his comments back in 2016-18

Quick new post: Auto-grading decade-old Hacker News discussions with hindsight I took all the 930 frontpage Hacker News article+discussion of December 2015 and asked the GPT 5.1 Thinking API to do an in-hindsight analysis to identify the most/least prescient comments. This took ~3 hours to vibe code and ~1 hour and $60 to run. The idea was sparked by the HN article yesterday where Gemini 3 was asked to hallucinate the HN front page one decade forward. More generally: 1. in-hindsight analysis has always fascinated me as a way to train your forward prediction model so reading the results is really interesting and 2. it's worth contemplating what it looks like when LLM megaminds of the future can do this kind of work a lot cheaper, faster and better. Every single bit of information you contribute to the internet can (and probably will be) scrutinized in great detail if it is "free". Hence also my earlier tweet from a while back - "be good, future LLMs are watching". Congrats to the top 10 accounts pcwalton, tptacek, paulmd, cstross, greglindahl, moxie, hannob, 0xcde4c3db, Manishearth, and johncolanduoni - GPT 5.1 Thinking found your comments to be the most insightful and prescient of all comments of HN in December of 2015. Links: - A lot more detail in my blog post karpathy.bearblog.dev/auto-grade-hn/ - GitHub repo of the project if you'd like to play github.com/karpathy/hn-ti… - The actual results pages for your reading pleasure karpathy.ai/hncapsule/

I've been asked: what do I think of @Kalshi & @Polymarket? These are still very early days. My vision, which I started to articulate ~1990, is of a world very different from both the world of then, and the world of today. A world where markets are accepted as offering more accurate estimates on far more useful topics. So I’m mostly interested in the potential of stuff today to enable and cause that future vision. The politics/policy questions on Polymarket and Kalshi trending now seem plausibly like topics where some people might find their estimate useful in making personal or collective choices. But I don’t have great confidence in that, or know who those people are. But if these systems continue to grow in size, and to attract users and competitors, they could lower the many costs of creating and managing such markets, allowing a lot more experimentation with markets like those I find more promising re my long term vision. Or course if these systems induce a backlash that gets them outlawed or drastically shrunk, that may plausibly block or at least long delay my vision. I don’t personally mind people having fun knowingly betting on sports, on actions that celebrities can influence, or on topics where insiders have big info advantages. But I see many people complaining about these things, and I fear a new prudish temperance movement may shut them down, and as a side effect shut down the more promising markets that I’ve envisioned. Robin Hanson

Implied leverage in binary options!

We're going to see the same market structure repeat in prediction markets that we saw in exchanges. Coinbase, Binance, Uniswap, etc. Multiple huge winners because of geography, regulation, and novel products. But the margins compressed faster on DEXes than on CEXes, which matters for what prediction markets can support.

Noice

gemini official release blog says it does 45.8% on HLE with tools. But the scale HLE leaderboard doesn't list the model version (deepthink or with tools). So, this question still trades at 40c while any reasonable retail user can misinterpret and buy these shares. The resolution criteria shouldn't have these many gotchas imo.