I've long been skeptical of the potential of prediction markets. A huge source of their recent growth, sports-related contracts, seems to rest on precarious legal grounds since none of the major players are registered sports betting outfits. There's a lot of activity around political betting, but these are concentrated around presidential and midterm election seasons. Is there really enough betting volume to justify an $8B valuation? What % of that volume will they be able to capture as fees? Over the past few years, the retail brokerage market has trended towards eliminating trading fees in favor of payment for order flow. Could the same happen to prediction markets?
Using valuation as a measuring stick is a bit silly given how much valuations can be inflated based on a pre-revenue idea and the pedigree of the company founder(s). When Mira Murati and Ilya Sutskever left OpenAI and walked into a room of investors saying, "I'm starting my own AI company", they arguably both created one-person, billion dollar companies right then and there.
The first batch of reviews for Vinfast's US vehicles were awful. I can't imagine this going well for them, though the stock will very volatile due to the low float.
I agree regarding leaving VC as an associate. Many of these positions are not partner-track. Also, in a world in which term sheets were being signed in 48 hours, it isn't clear what skills they've been building (no financial analysis, little-to-no due diligence, no deal sourcing).
Yeah, what skills transfer is there to a time of due dilligence that should mean something now versus being investor X in Wag’s Nth series raise because you can run digital shoulders with the big VCs and Ashton Kutcher…
Anecdotally, several CTOs I know intend to lessen their use of Scale, Labelbox and more in the future. Talked to one today who already ditched MTurk for GPT-4 -- cheaper, better, faster was what he said.
Labelbox does image annotating still, and one CTO said as soon as GPT-4 enabled this for him he'd have his team homebrew it from there.
Scale AI has long struck me as one of the worst examples of the excesses of the recent startup boom:
- An extraordinary amount of hype and handwaving over what is essentially a business process outsourcing operation
- Exploitation of gig economy workers in developing economies
- Significant data privacy concerns (this article describes how Scale AI data annotation workers openly shared amongst themselves private photos of Roomba customers taken without user consent: https://www.technologyreview.com/2022/12/19/1065306/roomba-i...)
I was a competitive chess player in my youth and am currently ~2300 on Lichess. Here are key aspects of any training regimen:
* Tactics - I did thousands of tactics puzzles using a program called CT-ART while under a time limit. I would have to complete 25 puzzles within 5 minutes while making 0 mistakes. If I made a mistake I would have to start from the beginning.
* Core positional concepts - Understanding pawn structure (double pawns, isolated pawns, passed pawns, pawn majorities, etc.), control of central squares, open files, king safety, and so on.
* Endgames - Many players neglect endgames. Master the basics: common rook-and-pawn and pawn-only endgames, how to checkmate with 2 bishops and bishop + knight, etc. Read Silman's Complete Endgame Course, Basic Chess Endings, and Dvoretsky's Endgame Manual (for advanced players) to improve your endgame play.
* Openings - Many people have mentioned learning openings, but it is generally a mistake for beginners and intermediate players to focus on this aspect of their game. It is not especially useful for a ~1000-rated OTB player to learn 15+ moves of theory in the King's Indian Defense only to end up in a position where there are slightly better according to theory but have no understanding of their position. It is better to center your opening repertoire around easy-to-learn openings and focus on improving other aspects of your game until you are strong enough to handle positions outside of opening theory.
The college student behind the account also tracks the private jets of other prominent individuals, including Bill Gates, Trump, Bezos, and a slew of Russian oligarchs. It should be noted that at the time I'm publishing this comment, other private jet tracking accounts (e.g. @CelebJets) remain active, suggesting that this policy of protecting privacy is being selectively applied.
"Having easy access to his whereabouts at all times" is hyperbole. Only his jet is being tracked, not his physical location at all times.
> ""Having easy access to his whereabouts at all times" is hyperbole. Only his jet is being tracked, not his physical location at all times. "
But knowing when he leaves a city and arrives at another is half the battle. A bad actor who wants to do him harm now know he just landed back in Austin. So if they were planning an attack, they could then try follow him from the airport, find him at home, Twitter HQ et cetera.
And for the record, to me this isn't about Elon. I don't support any of those accounts. I don't care if it's Brad Pitt, Bezos, POTUS, or my next door neighbor, it's Doxxing and it shouldn't be tolerated.
