Hacker News: marksimi

New comment by marksimi in "Advancing AI Benchmarking with Game Arena"

marksimi — Tue, 03 Feb 2026 02:13:43 +0000

The most popular form was solved in 2019: https://en.wikipedia.org/wiki/Pluribus_(poker_bot)

New comment by marksimi in "Meta chief AI scientist Yann LeCun plans to exit and launch own startup"

marksimi — Tue, 11 Nov 2025 17:59:32 +0000

Curious about how much risk Meta leadership was comfortable with when they decided to layer Yann. Perhaps the winds of open research were already blowing a different direction at the company, and he had already indicated that he wanted to leave as a result of that. We can only guess.

Kind of hilarious to me to consider him "failing" with LLMs. Given his remit was a research time horizon of 8-10 years, and the fact that he's gone on record saying that he expects the technology will stall out in the time horizon, it seems he can only take Ws and ties. Indirect influence on open-sourcing the models to propel research forward (which is pretty important for a chief scientist) which added benefit for Meta's other products.

New comment by marksimi in "Meta chief AI scientist Yann LeCun plans to exit and launch own startup"

marksimi — Tue, 11 Nov 2025 17:45:56 +0000

Yes, it was

New comment by marksimi in "Why Metaflow?"

marksimi — Sat, 16 Aug 2025 13:16:12 +0000

Curious to hear from folks who have used both Metaflow and Kubeflow to understand some of those tradeoffs.

Seems like Metaflow is comparatively lightweight, bit more tightly integrated with AWS, less end to end and a bit more agile.

New comment by marksimi in "How to teach your kids to play poker: Start with one card"

marksimi — Tue, 12 Aug 2025 13:00:15 +0000

Also used to be in that world and identify similarly in terms of my lack of love for gambling.

I'd suggest that you're empirically incorrect in saying that there is no perfect approach against a skilled player (6handed games which often reduce to a single heads-up interactions by showdown):

1. we know that a Nash equilibrium exists for every two-player zero-sum game such that it’s mathematically unexploitable

2. Pluribus approximated the Nash well enough (didn’t have to search over 10^161 possibilities) to crush high stakes skilled player over a good run of hands

New comment by marksimi in "How to teach your kids to play poker: Start with one card"

marksimi — Tue, 12 Aug 2025 12:44:05 +0000

You're not wrong that knowing the odds is a component of the skill, but to suggest that skill in poker stops there is minimizing many of the advanced aspects that require playing at a higher level (information management, assessing a player's likely range, determining the equity of a player's range with cards to come, realizing when your or their range is capped, etc)

New comment by marksimi in "How to teach your kids to play poker: Start with one card"

marksimi — Tue, 12 Aug 2025 12:29:14 +0000

I feel one of the most useful skills picked up by poker that people don't explicitly speak about is managing your information effectively.

Deceiving my opponent has the connotation of this happening in one instance. After you realize that you can't convincingly deceive your opponents in poker into perpetuity, it becomes a game of managing your image —revealing the right information while being conscious of information that you shared in the past (if you're playing someone skilled or perceptive, that is).

On the flip side, what an excellent game to help people pay attention to signals, figure out how to weigh them appropriately, and appropriately act on them when the situation calls for it.

New comment by marksimi in "Lossless LLM compression for efficient GPU inference via dynamic-length float"

marksimi — Fri, 25 Apr 2025 19:42:31 +0000

Time to (dynamically) float

New comment by marksimi in "Jane Street's Figgie card game"

marksimi — Sun, 16 Feb 2025 03:21:23 +0000

Real talk

New comment by marksimi in "Ask HN: How has AI changed your learning methods?"

marksimi — Tue, 25 Jun 2024 16:29:41 +0000

As someone focused on grad school, I find myself much less frequently getting stumped by problems and rage-quitting. I also use some prompts which help to speed up my learning in general while making sure any LLM doesn’t give me the answer.

One of my research interests is on how humans use expert systems (akin to how Go players’ ELO ramped significantly after the release of AlphaGo).

New comment by marksimi in "Ask HN: Taking a year off to go deep on ML/AI"

marksimi — Mon, 01 Apr 2024 08:36:36 +0000

I did this, mostly to start and wrap my MSCS (ML focus) and augment my data science skills. Would I have made the same decision in today’s hiring environment? Perhaps not. I was also solving for some other goals. Feel free to DM.

I’d wager that networking into a role would be best for most in your position given the market.

You might also consider trying to cut your downtime to allow for a retreat like the Recurse Center to take a focused leave while maintaining employment (still comes with risk).

New comment by marksimi in "An introduction to the theory and practice of poker (2020)"

marksimi — Thu, 22 Feb 2024 18:18:15 +0000

Definitely. And this isn't even bringing in what can be perceived as the "fuzzy math" of who has the range advantage, combinatorics, and consideration on how to proceed down the game tree on different streets (when the board changes texture).

There's certainly a part of me that wants to go thru this curriculum just to say that there's no way it could help someone to be better than an 'average player'.

Perhaps if the population on which the average is based is...the world?

New comment by marksimi in "An introduction to the theory and practice of poker (2020)"

marksimi — Thu, 22 Feb 2024 18:12:55 +0000

Yes, but my comment wasn't directed at that point. It was just that we can't reliably say that being a (theoretical) top 3 live game poker player allows us to isolate the most important (differentiable) skill at that level being live tells.

Further, I disagree with your point that the mathematics of poker (implicitly top tier highest level poker) are not complicated.

(edit) to your top line point: > My intuition having played recreationally is that the absolute optimal move in poker is relatively trivial to calculate compared to chess/go.

While the branching factor of Go is between 10^250 to 10^361 (and chess is ~35), poker is 10^17 to 10^165 depending on the game variant: ""Thus, the game tree for no-limit has a much larger branching factor and is significantly larger; there are 10^165 nodes in the game tree for no-limit, while there are around 10^17 nodes for limit (Johanson 2013)" - Sam Ganzfried, Reflections on the First Man Versus Machine No-Limit Texas Hold ’em Competition

...though it's the hidden information component which ratchets up the complexity and leads to a game tree size that's many orders of magnitude higher. If you're curious about that, there's some great info here on other aspects which add complexity: stack depth, multiplayer scenarios, etc: https://poker.cs.ualberta.ca/publications/billings.phd.pdf

That said, I do agree with a relaxed version of the point you're getting at: some subset of high level poker skills can generalizable quite well to other jobs.

New comment by marksimi in "An introduction to the theory and practice of poker (2020)"

marksimi — Thu, 22 Feb 2024 04:10:18 +0000

> are we not saying, essentially, that John is a human polygraph machine

No we are not. Poker skills (even at a high level) go beyond live tells.

New comment by marksimi in "An introduction to the theory and practice of poker (2020)"

marksimi — Thu, 22 Feb 2024 04:01:23 +0000

The paper referenced here is well beyond “just deep learning”.

New comment by marksimi in "Ask HN: Programming Courses for Experienced Coders?"

marksimi — Sat, 30 Dec 2023 15:53:51 +0000

OMSCS is ~$9k

New comment by marksimi in "Beej's Guide to Interprocess Communication"

marksimi — Mon, 27 Nov 2023 15:34:33 +0000

Came here to say the same. 100% not passing that class without Bern’s Guides.

New comment by marksimi in "Meta could become even more dominant in social media with Threads"

marksimi — Sun, 09 Jul 2023 20:36:59 +0000

Thanks for pointing out that oversight of mine. Bummer that I can't edit (past the window).

https://threads.net/@drstanier

https://threads.net/@msjewelia

https://threads.net/@kellan

https://threads.net/@louiebacaj

https://threads.net/@rands

https://threads.net/@mitraraman

https://threads.net/@neha.nark

https://threads.net/@peter_vukovic

https://threads.net/@s3nd1l

https://threads.net/@willrlarson

New comment by marksimi in "Meta could become even more dominant in social media with Threads"

marksimi — Sun, 09 Jul 2023 20:33:22 +0000

Few reasons: 1. I've gotten value from these folks on other platforms and mediums 2. I'm interested in what they have to share 3. I hope that starting to populate my following helps to choke out all of the bullshit brand recommendations I'm getting

Reason for sharing: can't imagine I'm the only one.

Not sure what you mean about the money thing. Whether you mean me or another person hypothetically following, I don't see a direct way to make money off of it.

New comment by marksimi in "Meta could become even more dominant in social media with Threads"

marksimi — Fri, 07 Jul 2023 18:34:27 +0000

It's early, but slowly getting more populated with people who post on Twitter. I wanted to pull together a list of Eng Mgmt Leaders to follow:

http://threads.net/@drstanier

http://threads.net/@msjewelia

http://threads.net/@kellan

http://threads.net/@louiebacaj

http://threads.net/@rands

http://threads.net/@mitraraman

http://threads.net/@neha.nark

http://threads.net/@peter_vukovic

http://threads.net/@s3nd1l

http://threads.net/@willrlarson