rgoldste

Reminds me of an encounter on BoardGameArena where the top ranked 7 Wonders Duel player complained there was a randomization bug (the Great Library never offered the science progress token). I thought he was raging (who hasn’t heard a poker player complain about bad luck) but turns out the developer checked the code and did in fact find this was a bug!

This is a plausible hypothesis. I’m curious whether OpenAI has considered this already and examined it I feel like an average senior eng could eval this in under two focused days, but maybe OpenAI has less unit-testing than I expect.

This is technically correct (the article notes this). Milk of all kinds comes from all cows. Some fraction of chocolate milk comes from brown cows.

Don’t feel like you need to understand 100%. You can always give yourself an hour to read a paper and gloss over some notation. If you read 5 papers over the course of a month, you can go back to your favorite and dive into the notation.

The truth is between these two. You can use a model’s output to train another model, but it has drawbacks, including model collapse.

It’s serious and actually tries to fix the company.

Very cool. Thanks for the details. Final question - how many rounds in a game?

Very cool. Some questions - How many games (?) in a row did you win and how much money was this? How many episodes did you appear on? How does this compare to other top players? Thanks.

Thanks for the response.

Curious, not trying to be a jerk. For something like ordering pizza (or even plane tickets), how much worse is calling to make the order as opposed to ordering online?

I didn’t quit fb but I unsubscribed from pretty much everybody. I think I’ve posted once in the last 6 years, though occasionally I come up tagged in friends’ pics. I think I’m better off. I occasionally miss engagements of my friends though so this is a downside. (I’m 29). I continue to use messenger to chat with friends, including one good friend who has a budget phone plan (meaning fb messages are free while texting this friend isn’t for her).

I guess I’m posting this here to say that there is a middle(?) ground that works for me. I rarely find myself envious or jealous on fb. I will confess that I do experience these emotions in my offline life (much less frequently than daily, more often than monthly). Come to think of it, while I spend under 20 minutes a week on LinkedIn, I feel some negative emotions while browsing this feed. I’ve just realized I should limit my LinkedIn better.

I’m not posting this to brag, just to share another perspective because it felt relevant.

Yes. Here’s a paper that dives into some of these points and a few others.

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2955455

For the uninformed - myself included - Who is this guy?

Is or was he ever affiliated with AWS?

Not disputing, but could you please provide a link or two that argues they are profitable in existing markets? I’ve only looked at their S-1, and there wasn’t hard information (e.g. single building operating statements) for me to conclude this.

I catch your drift, but this isn’t technically true. A company with negative discounted cash flows might have a positive expected value because it can undertake a risky project that has a negative expected value. Usually this project will bankrupt the company, but occasionally it will provide a windfall. During bankruptcy, the company might not pay off all of its debts, but it can’t be worth less than 0 due to limited liability. During the unlikely event of success, the company is worth something.

If you want to know more, look up “moral hazard” and “real options”.

To anyone who finds these models interesting and wants to learn a lot more - I’d highly recommend Mor Harchol-Balter’s textbook: https://www.amazon.com/Performance-Modeling-Design-Computer-.... (She is currently a professor at Carnegie Mellon).

(Self-repost from an earlier queuing theory thread).

Potentially relevant article (soft-paywall)

https://www.google.com/amp/s/www.nytimes.com/2019/01/11/tech...

To anyone who finds these models interesting and wants to learn a lot more - I’d highly recommend Mor Harchol-Balter’s textbook: https://www.amazon.com/Performance-Modeling-Design-Computer-... (She is currently a professor at Carnegie Mellon).