I worked at yahoo during its (in retrospect) decline.
It used to be hard to be "web scale" and available, now that's either k8s or a few checkboxes in AWS.
Yahoo used to be able to "coast" on the compellingness of their services because 80% attractive with 100% available and 100% global reach crushes 90% attractive with 95% available and 25% global reach.
I was often confused by the hyperfocus of analysts asking "Is Y! a tech company or a content company?"
What they were really asking was if we should be valuing Yahoo! as 30%+ margin on putting ads next to Yahoo! News articles, or 10x multiplier on originating GMail/Search?
I think "data is the only moat", and in a way that goes back to the "first to market / eBay" POV, and the difference between first to market and fast follower is super interesting!
Very true, but you're a decade or two late for that. IIRC Cash4Clunkers put like a $3k floor on used car value (~$5-7k in today's dollars) meaning you'd never sell your old car for $2k to an individual when you could sell it to the government for $3k.
Per google it was started in 2009, which means any car worth less than $5k around 17 years ago isn't materially impacting new or used car prices today.
...but seriously... there was the "up until 1850" LLM or whatever... can we make an "up until 1920 => 1990 [pre-internet] => present day" and then keep prodding the "older ones" until they "invent their way" to the newer years?
We knew more in 1920 than we did in 1850, but can a "thinking machine" of 1850-knowledge invent 1860's knowledge via infinite monkeys theorem/practice?
The same way that in 2025/2026, Knuth has just invented his way to 2027-knowledge with this paper/observation/finding? If I only had a beowulf cluster of these things... ;-)
I want to pitch to my local makerspace "log-10" manufacturing.
Basically there's a ton of traction at the zero-to-one (making the first prototype) and then you start looking at how to "scale" your manufacturing (ie: making 10 at a whack), and then eventually you MAY get to building/assembling 100 at a whack, and up to 1000's or more (where you'd "graduate" to partnering with a "real" manufacturer).
Maybe it's just the way that I'm wired, but I've done 3-4 projects where I've gone down the B.O.M. rabbit hole and have scaled to at least 100 assembled/packaged items.
It seems like a local makerspace is the perfect launch-pad for having flexible "staff" (ie: other makerspace members) that can handle ambiguity and would be invested in the success of a locally owned/managed product!
"""under perfect play all chess games be a the same single one outcome of the following (we just currently don’t know which one, “A” playing the white pieces):
Mr. A says, “I resign” or Mr. B says, “I resign” or Mr. A says, “I offer a draw,” and Mr. B replies, “I accept.” That is, under perfect play, each chess position is either a forced win, forced draw, or forced loss. The domain of a perfect chess position evaluation function is these three cases as symbols."""
...basically if you think you're going to win (aka: you have a 200 centi-pawn advantage), you can offer the doubling cube to your opponent (doubling the stakes of losing). If you're playing to win $5, and halfway through you think "yep, 90% chance I'm going to win this one...", you push the doubling cube to 2x (aka: $10 consequence), and kindof like poker your opponent has to evaluate whether it's "worth it" for them to stay in the game.
You might imagine a "2xELO penalty" where White takes a Queen with a Pawn, and then offers "2x, or I'm gonna beat 'ya!". If Black say "Naaah, you just activated my trap card!" and then either accepts "2x" or pushes back at "4x", then it becomes a little more like poker... you think you can beat me, then prove it!
Not that I'm suggesting changing the rules of Chess, but overall I'm really fascinated by the concept of formalized semi-out-of-band risk-taking to potentially end games early.
The doubling cube works well in Backgammon because it is a rare example of a popular game with randomness, without hidden information (every information set contains exactly one node of the decision tree, if you want to get extremely technical,) and, critically, with "different endings" (normal win, gammon, backgammon.) Doubling decisions are especially interesting because while they're always objective (it could never be the case that perfect players disagree on the correct move, that requires nontrivial information sets,) it could be the case that:
- it's correct for a player to double and for the other to accept;
- it's correct for a player to double and for the other not to accept;
- the position is "too good to double," because the equity from the probability of a double or triple game exceeds the advantage you'd get from a double;
- all of the above being influenced by the match score, e.g. if I'm 3 points away from winning and you're 5 points away from winning, I could make different decisions than if it were the opposite.
Chess has none of them, the doubling cube would be exclusively a psychological power play, something like "it's theoretically drawn but I don't think you can defend it," which is not a great game dynamic.
In general, transplanting the doubling mechanic without a similarly rich context doesn't tend to work well.
Games like backgammon (that have betting and the doubling cube to continue), Go (which is calculated in stones), and bridge (again having points) have more natural intermediate scoring systems than chess.
In my opinion the "winner takes all" aspect of chess is similar to what makes analyzing voting systems difficult. In a non game context: Aspnes, Beigel, Furst, and Rudich had some amazing work on how all or nothing calculation really changes things: https://www.cs.yale.edu/homes/aspnes/papers/stoc91voting.pdf .
For a while I really dug in to multiple player (and teams-of-players) ELO calculations. I got into an argument with my friend about whether second place was any better than last place... specifically in poker, but applicable to multiple games (imagine chinese checkers [race to finish], or carcassonne/ticket-to-ride [semi-hidden scoring until the end]).
His POV was that "if you don't win, you lose" and my POV was "second place is better than last place". His response was: "if I play poker to get first place it's wildly different than playing for second or third place [and I may end up in last place wildly more often due to risk % or bad beats]"
I've been more used to "climbing" type performance games (ie: last place => mid-field => second place => first place) and in my gut I wanted my ELO to reflect that (top-half players are better than bottom-half players), however his very valid point was that different games have different payout matrices (eg: poker is often "top-3 payout", and first may be 10x second or third).
I think in my mind I've settled on EV-payout for multiplayer games should match the "game payout", and that maybe my gut is telling me the difference between "Casual ELO" (aka: top-half > bottom-half), and "Competitive ELO" (aka: only the winner gets paid).
Go is also winner take all. It's psychologically satisfying to have a big win, in the same way that it's psychologically satisfying to achieve a brilliant checkmate, but in any ordinary game or tournament (outside of certain gambling setups), a win by 1/2 point is the same as a win by 20+ points.
Yes and no. One could say this of any game with points where the margin of victory doesn't affect long-term outcomes (e.g. most ball games).
A win by 1/2 point or 20 points it suggests a very different relative skill between the two players. Similarly the custom of the stronger player playing white without komi suggests that the point differential matters.
I see what you're saying; this is true for any game scored win/loss. Even gridiron football if you're down by 4 points with time almost out you won't kick a field goal (worth 3 points).
I'd like to point out that some online chess tournaments, mostly using rapid and bullet times, have a "berserk" option pre-start, where the player taking it halves their allotted time bank, for double the winning points.
It's not a bluff, since information is still 100% open to both players, but it changes dynamic a lot.
I've always thought of it more as "Co-Pilot", but formally: "Autopilot" might truly be the better definition (lane-keeping, distance-keeping), whereas a "Co-Pilot" (in aviation) implies more active control, ie: pulling you up from a nose dive.
So... informally, "Tesla Co-Pilot" => "You're still the pilot but you have a helper", vs "Tesla Autopilot" => "Whelp, guess I can wash my hands and walk away b/c it's AuToMaTiC!"
...it's tough messaging for sure, especially putting these powertools into peoples hands with no formal training required. Woulda-coulda-shoulda, similar to the 737MAX crashes, should "pilots" of Teslas required training in the safety and navigation systems before they were "licensed" to use them?
...you need "man" pages for moderately-comprehensive options explanations (backed by /usr/share/doc/$TOOL/README.txt if you're a debian user), but "bro" tends to focus on the "yo, this is what you're actually trying to do here...", including sometimes crossing traditional "this-command" boundaries (eg: in the diff example, offering `diff <( cmd1 ) <( cmd2 )` b/c sometimes that's what people are trying to do).
I can't find one that I submitted but it was something like `bro sed` => `# bro, just use awk! => awk -- '{...}'` ...basically you could go down the wrong rabbit hole, and there's kindof a nice little community of users helping to lift each other up (with upvote/downvote) and focusing on providing relatively simple and salient examples rather than a wall-of-text-options where you know that it's possible, but you don't know how to start. (eg: see `bro ffmpeg`)
It used to be hard to be "web scale" and available, now that's either k8s or a few checkboxes in AWS.
Yahoo used to be able to "coast" on the compellingness of their services because 80% attractive with 100% available and 100% global reach crushes 90% attractive with 95% available and 25% global reach.
I was often confused by the hyperfocus of analysts asking "Is Y! a tech company or a content company?"
What they were really asking was if we should be valuing Yahoo! as 30%+ margin on putting ads next to Yahoo! News articles, or 10x multiplier on originating GMail/Search?
I think "data is the only moat", and in a way that goes back to the "first to market / eBay" POV, and the difference between first to market and fast follower is super interesting!
reply