Thousands play one game of Pokemon simutaneously | Hacker News Reader

Thousands play one game of Pokemon simutaneously

twitch.tv

28 pointscrystalmace24 comments

It would be cool to see this done with a voting system, rather than the current system which (because of how the actual game handles input) more or less randomly selects a button to press. If, instead, it looked at the buttons pressed in the last n seconds, and then picked the one with the most occurrences, I think you'd get more of a hive-mind effect.

That is the best suggestion I have heard so far. That would make the game really interesting and everyone would get to play.

Currently most of the input is lost, because even though the system processes every one and the game emulation receives everything, the game is still ignoring it on animation, menu transitions and so on.

I hope you don't mind if I message the profile with the idea.

Go for it.

Doesn't it already accomplish this? The game accepts an input every n frames; if the chat has a higher occurrence of a certain key input it has a higher chance of accepting that one.

No. Now there is a chance that the key with the highest occurrence is selected, but in the method proposed the highest key would get always get selected.

So, right now it's anarchy. You're proposing democracy. How long until this turns into a dictatorship ruled by the players who know what's the best way to play? :)

However long it takes until one of the payers get frustrated enough to just play it by themself on a emulator on their own computer. :)

Right now it's actually still a sort of democracy; it's just that the voting system is extremely noisy.

That's exactly the idea I had when I saw it!

Yeah would be interesting to have a form like strawpoll that users voted on and every 20 or 30 seconds a new command was input

This would be interesting to do, and to play with the time given for each voting block. Depending on the number of people playing, I doubt you'd need very long to get a pretty good sample for the next move.

Either way; it'd be definitely be neat

Interesting results and fan art: https://docs.google.com/document/d/1Ka89uCBeLl9VV5GPlAKuW7yF...

And here come the imitators:

http://www.twitch.tv/xkeeper_

http://www.twitch.tv/trakof

http://www.twitch.tv/piapwns

That fan-art / random section is hilarious, thanks!

Well, it's been fun watching Ash fail to get past a door for the last 15 minutes (EDIT: it looks like it's been stuck there for 4 hours). There's some people obviously trolling.

I guess this experiment shows collective control doesn't get too far, objectively, without some sort of selection (e.g. punishment for bad moves).

Punishment is lack of progress; the longer they are stuck more people will want to progress, and those who oppose it will be a minority.

They almost made it over the "bridge" a couple of times, they were two steps away from a safe location once.

It looks like the player has been stuck on a part of the map that is easily sidetracked by one or two "down" commands.

Here is a graph someone posted in the chat: http://i.imgur.com/6Iy7h7l.png

as suspected nothing is happening, thousands of people are all giving conflicting commands so the character is just wandering aimlessly without really going anywhere

I'm quite astonished that they have managed to make it past three gyms already. I suspect that the game will eventually be beaten, but only after a while and after going through several ' cycles', so to speak.

I imagine that each cycle will consist of a mass of people becoming bored with the lack of progress and leave. As more people leave, it becomes easier to control the character, leading to progress being made. Howevere, once progress is made and certain milestones are reached, the interest will again grow and the players will increase again. Eventually, it will become too dfficult to play and interest will wane again; allowing the cycle to repeat.

Thus, I think it will eventually be completed, but not for a while. I'll be following it with interest however!

Very nice experiment that really proved worthwhile to the developer. With some luck and media coverage, they have enough viewers now that they have introduced the subscribe (read: $) option, not present before.

I would really like to see more games done in similar fashion.

I'm wondering about the legality, but also the latency. How long does it take to register an action?

From the info below the video:

> The amount of lag is approximately 20~40 seconds depending on connection quality. After watching the stream for awhile the lag may increase a tiny amount (enough for chat spoilers), refreshing may help in this case.

It's streaming lag, the input lag should be far higher.

I suspect a very simple setup :

- An IRC bot listen for message including a key name and queue them somewhere (Redis maybe?).

- Another process just dequeue keys and apply them in emulator.

It says in the description( for teh lazy ~ 20s-40s ). I have experienced about 20 seconds.

EDIT: Actually, the time to register the "keypress" is almost zero, but the effect is only seen many seconds later. You have to guess what the game state will be in ~20s-40s.

The input rate us pretty crazy right now, the game is barely playable. There was some slow constant progress when there were about 4000 viewers, and they even managed to catch a few pokemon when the view count was low.

LOL, very funny.

Reply on news.ycombinator.com