131–140 of 153 posts
If a game can be easily played by a computer maybe it's time to start playing a more complicated game, worthy of your human brain
Name one game that a human is better at than a trained NN. Starcraft 2, chess, go, Dota 2, have all been shown that AI wins. If a game is digital, AI will win after sufficient modeling. I don't think any of that negates the merit of the games themselves for human play.
Earlier quoted context omitted.
Just make a "usb mouse" which is actually a programmable arduino? What makes an aimbot possible is that something can read the game's state and turn it into a mousemove signal. It can then feed it into the arduino instead of directly moving the mouse. The problem is preventing game state from being read. Part of this can be done by detecting DLL injections. Sadly you will always be able to write a monitor that runs a…
Any fake USB mouse that someone makes would not have the Intel Secure Enclave that signs mouse movements. So the mouse inputs would be rejected by the server/banned.
Earlier quoted context omitted.
Any fake USB mouse that someone makes would not have the Intel Secure Enclave that signs mouse movements. So the mouse inputs would be rejected by the server/banned.
Why wouldn't it? How would the other mouse manufacturers get it?
This is close to my ideal way of avoiding a runaway Ultron/Skynet-type AI scenario: airgap it and force the use of a physical mouse, monitor, and keyboard to interact with the outside world. At least once it reaches Twitter we can stop it before it destroys humanity one robotic keystroke at a time.
Earlier quoted context omitted.
I don't fraternise with those filthy LoL players! /s On a more serious note, I'm sure something like this exists already. I wasn't even the first one to write a Dota picker, I got the idea from someone else's tool. I even saw one that could analyse the display and recognise the hero images to enter the existing picks automatically! Even my little utility merely scratched the surface. It didn't handle the first few pi…
When it comes to solving complex AI problems I always like to think, if I had given a human this task, what would they want to know. And I think even a human would struggle winning the character select mini game just with tabular data related to the players' win rates with those characters in previous matches. Learning about actual play styles of each playerand customising their play style with a given character woul…
In retrospect, a neural net would have likely been the best approach, but the (maximal!) noise in the training data would have made it difficult to train. I wanted a billion or more game outcome records to enable NN training, but the game stats API is very throttled and it would have taken months to get that much data. In that timeframe the game is often patched with new rules, which would invalidate the older data.
Earlier quoted context omitted.
When it comes to solving complex AI problems I always like to think, if I had given a human this task, what would they want to know. And I think even a human would struggle winning the character select mini game just with tabular data related to the players' win rates with those characters in previous matches. Learning about actual play styles of each playerand customising their play style with a given character woul…
Precisely. The top-10% thing essentially made my tool into an "expert system". I built a simple model with a bunch of parameters, and had it figure out those parameters based on observations of high-skill games. In retrospect, a neural net would have likely been the best approach, but the (maximal!) noise in the training data would have made it difficult to train. I wanted a billion or more game outcome records to en…
Earlier quoted context omitted.
With you on progression. First time you see someone being super fast you call them a cheater. Then you learn about Strafe jumping and that it's just how q3 physics work (even if unintentional from the programmers perhaps) and you learn it suddenly you're called a cheater by n00bs. Same with jumping onto things of a certain height. If your graphics hardware could not do a certain framerate constantly it was impossible…
> Having played q3 back when myself it was standard though to make _all_ skins be the standard blue skin. This must’ve been early on in q3’s life, ‘keel/fb’ (bright green Keel) has been the preferred enemymodel for quite some time since it’s highly visible, has loud movement sounds, and fills out the hit box pretty well. I always assume my opponent forces fullbright Keel for my model too.
Earlier quoted context omitted.
Precisely. The top-10% thing essentially made my tool into an "expert system". I built a simple model with a bunch of parameters, and had it figure out those parameters based on observations of high-skill games. In retrospect, a neural net would have likely been the best approach, but the (maximal!) noise in the training data would have made it difficult to train. I wanted a billion or more game outcome records to en…
Makes me think you could have the character stats and game design decisions as parameters, and have it select characters based on attributes rather than hit encoding of their simple identities (names) but from what I know of LoL the unique thing is character abilities aren't really differentiated in that modular way and the exact values it aren't given (like a spin attack movie's exact radius, though I bet that could…
An interesting effect I noticed was that the pub games have all sorts of weird second-order effects.
E.g.: Some characters are unpopular because pub players play them ineffectively, and/or they're hard to play well because they're so niche, so few people get enough practice with them in the right kind of scenario.
Dota's Jakiro is a great example. It's a lumbering support hero with very slow spells and a glacial turn rate. It's like trying to do acrobatics with a jumbo jet. He's frustrating to play and when most people pick him, the effect on the overall win rate is about -15%, which is insanely bad for 1 out of 10 players in a game! My win rate with him was something like 25%, which is absolutely atrocious. That's "throwing the game" bad.
My picker app often recommended Jakiro when the enemy team had over-represented heroes that typically commit to a fight, such as Legion Commander or Axe. Those heroes are nailed to the ground in a fight and can't escape Jakiro's devastating-but-slow attacks.
I ended up playing Jakiro a lot, and eventually I got about a 55% overall win rate with him, which is an amazing swing if you think about it. It took practice though, figuring out how to best utilise him. If I had played him in random games, I never would have gotten anywhere. The picker app however made this possible, by providing these hints of when and how to use the hero.
Earlier quoted context omitted.
Name one game that a human is better at than a trained NN. Starcraft 2, chess, go, Dota 2, have all been shown that AI wins. If a game is digital, AI will win after sufficient modeling. I don't think any of that negates the merit of the games themselves for human play.
Any 4X game.