At the end of the game, yes, but if players don't try to bankrupt each other it probably won't happen.
Early in the game the expected value of a trip around the board is positive because of low rents of undeveloped properties and the net gain from Chance and Community Chest, plus passing Go. Cash flow into the game is positive as a whole.
Later in the game, when property is developed you're going to be paying bigger rents. This is going to cause players to mortgage or sell houses to pay the big rents (also their own liquid cash supply is likely lower from developing their own properties), each of which is a money sink since you don't get full value for these. This causes money to exit the game, eventually leading toward bankruptcy.
However, this requires that players actually get to develop their properties. If the game hits a lock due to no one trading and no one getting a natural monopoly, the game can go pretty much indefinitely.
> If the game hits a lock due to no one trading and no one getting a natural monopoly, the game can go pretty much indefinitely.
Yeah, I once played with someone whose strategy was "Never trade". In an act of desperation, I offered him $1,000 for a light-blue property, and he refused. The game lasted for nearly an hour before I quit out of extreme boredom.
A typical 4-player game of Monopoly should only last around 30 minutes, but people often have terrible strategies or play with ridiculous house rules that constantly inject money into the game, prolonging it. A lot of people don't even know that the house rule they're playing is even a house rule because it's what they were taught when they first learned to play.
> Yeah, I once played with someone whose strategy was "Never trade".
That's the right strategy for strong play. You should never trade property for mere money. If you absolutely have to, trade property for property if the trade gives you a monopoly, or if (in a more-than-two-player game) doing so gives you a net increase in the number of groups for which you hold at least one property (so you can block anyone from getting a monopoly in those groups without trading with you).
It's not an especially fun strategy, either for you or the people you're playing with.
The discovery was awful too. They had a similar exclusivity deal for some Dota 2 tournaments, and I could not find the stream from the Facebook home page while it was live.
Youtube has a similar problem, It's almost impossible to figure out who the hell is actually streaming a specific game, search for streams, or even when the people you are subscribed to are streaming.
I've heard this was a problem since the spice level was optimized in an office where they make the cookie dough the day of service, but deployed in an office where they make the dough the night before and cook on demand. The spices in the dough longer made the flavor stronger than it was optimized for.
- The 1v1 bot played at The International used a special creep block reward (and a big if statement separating that part of the agent from the self-play trained part). It trained for two weeks.
- A 2v2 bot discovered creep blocking on its own, no special reward. It trained for four weeks.
- OpenAI Five does not have a creep blocking reward, but neither (to our knowledge) does it creep block currently. Trained for 19 days!
I see. Thanks! So it manages to win lanes without even creep blocking? That's quite good. Any chance you could share the last hits @ 10 mins for the games it has played (for both bots and humans)? I think that's a crucial number to judge how OpenAI Five is winning its games.
I believe the article said that Blitz rated the bot last-hitting at about average for humans, although he might over-rate what an average human player last hits like.
Yeah, he might be overestimating 2.5k mmr players, and there's also something to be said about the consistency by which the bot last hits. A human player would have a high variance of last-hit performance, while the bot will probably guarantee a minimum amount, thus ensuring a minimum set of items needed for the mid-game transition.
But my larger point is, the early game doesn't have a lot of strategic elements in it. You have to last hit, not die, harass opponent, get items. You can play it by the book pretty much. The challenge in early game is to be able to handle 5 different things at the same time. So there's never really a question of what to do, but doing it does require mechanical prowess, which we know bots can easily be better at, than humans.
The team composition chosen is very early game snowball oriented. So is the bot winning simply due to mechanical superiority and early game advantage? Access to last hits @ 10 mins, gold and net worth graphs would allow us to answer that question.
Second, we can bring what is known as differentiated education — based on the idea that students master skills in different ways and at different speeds — to every student in the country. A 2013 study by the National Institutes of Health found that nearly 40 percent of medical students held a strong preference for one mode of learning: Some were listeners; others were visual learners; still others learned best by doing.
This study was a 100 person survey with no experimental evidence to back up people's beliefs. I've been working with educators for a while, and they've said that "learning styles" are just a myth with no solid evidence. All this survey says is that people think they learn best via one method, but people are notoriously bad at knowing what is an effective way of learning. In surveys people overvalue the "intensity" of a learning experience [1], which makes them think things like a 3-week bootcamp are more effective, when really spaced repetition is a better use of time.
At least in the US, there are many easier ways we could improve our education methods before resorting to ML-driven customization.
I don't really have experimental data to back this up. But after struggling for my entire life until recently with learning this is the experience I have had studying for my last course(after a decade of struggling to study).
I use to write everything down, it was the only way to retain whatever little information that I was capable of retaining. I never had enough confidence to NOT write everything. Realizing that all my abilities were impaired because of improper cognitive development. I decided to start from scratch, having nothing developed and picking one area that I was suited to. I have been practicing reciting rap lyrics of Eminem and Nas(not really for any reason, it was just fun). What I realized is that now my ability to recall information was significantly improved at least from a verbal learning style perspective. And it only took a few months develop.
While studying for my final I went back to my old habits of writing every single detail down and realizing I will never have enough time to write it all down. So I started reciting knowledge the same way I would recite rap lyrics and I retained EVERYTHING. I have never studied so little, yet I aced my exam got an A in the course and was proud.
Moral of the story is that people are different, and while most people may learn from one method, it doesn't mean they will have that capability developed at the time they need it. So they have to learn to adapt with what is available to them.
edit: I will add that again poverty is the source of all this. My parents grew up in severe poverty and never were able to develop these skills properly and I had to suffer because of that.
Very complicated issues, but mostly traumatic childhood of my parents which was mirrored by my brain and had to be undone before any serious cognitive development could happen. Trauma = Disassociation = Brain functions become disconnected and do not work properly. It becomes harder when you have to bootstrap your brain to work well and you are basically starting over 20 years later than everyone.
>I've been working with educators for a while, and they've said that "learning styles" are just a myth with no solid evidence
I've worked with educators too. They are and they aren't it depends on the level and time period you're taking. At the individual level how a math problem (for example) is presented can affect how quickly a student grasps the concept. When you're taking an entire school and more than a decade of schooling the effects of presentation aren't all that meaningful because people do the best job retaining the knowledge they use frequently.
I believe the comment you are responding to would say "both".
The learning styles to which they refer is the notion that different people learn better through visual or audio or hands-on experience. Spaced repetition versus intensive study is a different question, one of practice and time use, that doesn't address the question or value of "style".
At least, that's how I understand it. As a casual interpreter of someone else's comment, I could of course be wildly wrong.
My understanding is that there are two parts to this suit: patent infringement and trade secrets. You can't patent a trade secret (then it wouldn't be secret), so these are two independent things.
Alsup was suggesting Waymo to drop the patent portion of their suit and focus solely on the trade secret portion of their case. The trade secrets were pared down as mentioned in that snippet, but that's in addition to dropping 3 of the 4 patent infringement claims.
I disagree with this. In my experience, improving your dataset quality (either via gathering more examples, getting more reliable labels, or doing better normalization) has yielded bigger gains than trying to use the latest and greatest model architecture for the problem. Linear and tree models are still widely used in applied problems.
That post doesn't really make sense. Of course if you can get two Monopolies without any other player having a threat you will win. Building shortages are rarely used in serious play.
The author admits as much in reply to Minus-Celsius. He wasn't proposing the master strategy to win against good players, he's laying out an effective way to poison the well so we can stop playing that stupid game every thanksgiving.
Early in the game the expected value of a trip around the board is positive because of low rents of undeveloped properties and the net gain from Chance and Community Chest, plus passing Go. Cash flow into the game is positive as a whole.
Later in the game, when property is developed you're going to be paying bigger rents. This is going to cause players to mortgage or sell houses to pay the big rents (also their own liquid cash supply is likely lower from developing their own properties), each of which is a money sink since you don't get full value for these. This causes money to exit the game, eventually leading toward bankruptcy.
However, this requires that players actually get to develop their properties. If the game hits a lock due to no one trading and no one getting a natural monopoly, the game can go pretty much indefinitely.