Hacker News new | comments | show | ask | jobs | submit login
Game Theory: How 70,000 Pokemon Players Sabotage Themselves (minimaxir.com)
289 points by minimaxir 1393 days ago | hide | past | web | favorite | 69 comments

This is a fascinating example of the power of biased randomness. The gameplay challenges this article points out are notable because they require sequences of inputs that are both specific and lengthy. Even if the input stream were random, there would still be a nonzero probability that the correct sequence of instructions would be realized.

On the other hand, this stream isn't random. If it were truly random, the player would just move pointlessly in a horrible Brownian motion. It's nonsensical, to be sure, but in some weird way it encapsulates knowledge about the game, and as a result the game makes progress.

It sort of enlightens other places where true randomness is required, and the presence of any information or understanding radically changes the behavior of a system. In cryptography, even the slightest weakness in the probabilistic underpinnings of a cryptosystem can render it useless. In finance, even the slightest edge over the market can be leveraged to produce gains.

You might be interested in this: http://www.twitch.tv/RngPlaysPokemon

An actually randomly (well, pseudo-randomly) controlled Pokémon game.

Interestingly it's currently on its second attempt. The first attempt was deemed a failure after earning the second gym badge, when it got stuck in the wrong town and released its starter Pokemon, leaving it only with a single Magikarp it had purchased from an NPC. (I guess it's pretty hard to catch a wild Pokemon using random button presses.)

To quote someone from the subreddit: "The problem wasn't having a magikarp, he could still level it and get gyarados. The problem was not being able to EVER have a pokemon that can learn cut. He was out of money to get pokeballs, and out of trainers to get money from defeating."

This wouldn't have been a problem in Generation 2 and onwards, because there you can exchange numbers with trainers to have rematches which can yield more money.

How can you level a magikarp without being able to get another pokemon to actually win fights? I don't think you get anything from having Abra flee...

Once it's run out of PP it can win fights with Struggle.

I watched last night as they went back to the PC just to retrieve the Helix Fossil. That was madness.

The mythologizing of the random crap that happens in the game is the best part. I like how opposing cults are being built around different items/characters in the game. It's more interesting than the fact that things actually get accomplished with the seemingly random actions.

The article's URL is "Glory to the Helix", and there are derivatives like "Ask the almighty Helix Fossil"


Hello, sir. Do you have time this morning to talk about our Lord and Savior, Bird Jesus?

To add to that, I believe the goal there was depositing the 'dig' rattata that they kept escaping the dungeon with, finally fed up enough to take the risk.

And then the process of smashing random buttons at the PC restored the precious helix fossil.

And the players decided to accept this wonderful result and not tempt fate, and left to continue on their journey.

Still with the rattata.

I don't like watching others play games (never did) but I find reading about things like the Eve battles, strange things that happened while testing engines (Oblivion), and this after (or during) the fact.

This one I just love reading about. I think it's because of the way people write about it. It almost sounds like Hitchhiker's Guide or something the way it's so random but accurate.

Trying to explain this to people in person has resulted in some odd looks but I still try...

I totally agree, things like opening the AQ gates in WoW are like old battle stories

There is no goal. There is some coordination in the irc channel but otherwise we're just assigning narratives to mostly random events^1. And that's beautiful about it.

^1 Which we actually do constantly, i.e. athletes being mostly weighted random number generators, but at least then there is a thin veneer of motivations at play.

What actually happened upon accessing the PC was almost entirely random, of course.

But going to the PC vs. staying away from the PC is very much a goal-based decision. A fight between people that want to change the pokemon/items being carried and the people that are unwilling to take that risk, with majority rule.

My god.


the developing fandom alone makes it worth watching


Check this out. You can clearly see it the majority of the commands move in the desired direction (or at least some of the time).

It's not the fighting and trolling that causes 90% of the problems, it's the delay on the video that means everyone has to guess where the game will be half a minute in the future.

Such a shame that twitch changed streaming technologies recently. It used to be easy to get as low as 2-3 seconds of latency. A world of difference in something like this.

For example, that ledge, it was easy to get enough 'right' movement to overpower malicious 'down' commands, because the first input in a direction only turns the character. But there were 'down' commands needed to get to the ledge, and the lag caused them to keep pouring in after they were no longer needed.

This article is a nice overview of the spectacle but its premise is fundamentally flawed.

Twitch did this in response to a huge amount of complaints about lag. It reduces the chance of stutter lag, but hinders audience interaction. They figured it was worth the trade. From my experience, they do get a lot less huge complaint threads on reddit and related gaming sites.

Some of the trolling isn't trolling, it's a tactic -- specifically the constant start spamming.

Since twitch.tv recently added up to 30 seconds of stream lag, you want to spam start during the trickiest movement sections to minimize latency between the stream and gamestate. This is most important in ledges and mazes!

More of the chat catches up to the current position and starts putting in the right input, which actually has a higher chance of being accepted, after the start spam delay.

I think what you mean is "some of the constant command spamming isn't trolling, it's a tactic". Since trolling tends to be defined by intent, if the intent is honest, it's usually not called trolling.

Yup, or some of the so-called trolling isn't trolling.

I was not under the impression that the article author called that trolling. I understood that there would be enormous difficulty in coordinating the inputs among the honest players even in the absence of trolls.

Not much on game theory here; the article was mostly a summary of what happened so far on stream.

Exactly. This is game theory how? What is the best response? What is the equilibrium concept? If there isn't an answer to these questions, it's not game theory.

Would be nice of you to credit the artists whose work you rehosted, minimaxir.

The millennial generation's version of the Infinite Monkey Theorem[0]?

[0] https://en.wikipedia.org/wiki/Infinite_monkey_theorem

Here is the video of ABBBBBBK( and JLVWNNOOOO being released by trolls: http://www.twitch.tv/twitchplayspokemon/c/3738870

It's a game, not everyone is playing it for the same reason and the input is anything but random. There aren't really any conclusions you can draw from this other than it's entertaining.

I'm not sure if there conclusions that one can draw from it, but I think it has use for reasoning about security - both physical and virtual. It sounds like the people who want to progress in the game have started to adopt a set of "best practices" to both deal with honest players and trolls. I think the constant presence of trolls makes it quite interesting.

The obvious parallels are to a public-facing computer system, where attackers are constantly trying to get in. The difference, here, though, is that no one has privileged access. I find that an interesting thought experiment: is it possible to have any security when attackers have the same rights as legit users?

Except without a rigid definition of security, or goals there's nothing to compare, analyze, hypothesize, or even observe.

And that question, everyone already has equal rights on the internet, that's why security software focuses mostly on identity verification.

You don't need explicit things "to compare, analyze, hypothesize or even observe" in order to influence how you think about a problem.

And on individual places in the internet, people do not have equal privileges. For example, there are things that the moderators here on HN can do that you and I cannot.

If you don't define your problem, how can you come to understand how you think about it.

That's achieved through identity recognition. All computers are generally treated the same.

>is it possible to have any security when attackers have the same rights as legit users?

Isn't this roughly the problem that bitcoin set out to solve?

This so fascinating, I am almost certain that a new genre of gaming experience has been created and people will demand more games like this one! This is complete madness!

Well this is definitely a new take on "social," gaming. Instead of fighting each other, work together in an uncoordinated manner.

I love to see new implementations in gaming, I feel the gaming industry really cultivated me as a hacker, and Pokemon was my first hack....anyone remember getting "Missing No."

I wonder how long it would take if the game went round robin. Everyone in the chat got a chance to enter a move and the move was guaranteed to work and everyone saw the outcome before the next person gets to press their button. I predict much faster.

What about something more democratic? After a button is input, wait until a button has been said in chat N times (for some N>1) before pressing that button. It wouldn't solve the other problems, but it would make it so trolls wouldn't be able to disrupt gameplay as much unless they have a majority.

Oh, that's how it works right now, with a twist. There's "anarchy" and "democracy" modes. To quote the channel readme:

    TwitchPlaysPokemon now has two modes, anarchy and democracy.

    Anarchy mode is the "old" mode, where everyone's inputs are applied immediately.

    Democracy mode is vote-based and has a more sophisticated input system.

    In order to switch from one mode to the other, the mode that isn't active needs 75% of votes as indicated by the dotted line, the current percentage of votes is indicated by the black line.

    In democracy mode you can compile a sequence of inputs. left2 will move left twice, left2down2 will move left twice and down twice.
The "dotted" and "solid" line referred to are two indicators moving in a progress bar between "democracy" and "anarchy"

That was implemented earlier today. Not quite like that but as a voting system to let the most voted command get inputted after 5-10 seconds. They can also vote to switch between modes which I think has definitely biased things to being a little less random, but it's made some users angry and had them voting "start9" to halt all progress by hitting the start button 9 times.

The streamer tried that. It would poll for a few seconds, then execute the majority command. A large number of people opposed this and began to spam start in protest, and actually killed all progress until the streamer reverted it. At the moment there's a slider that allows the chat to vote between 'democracy' and 'anarchy'.

well...that does give them something to spam for instead of start :)

I'm still amazed that anything was even achieved in that game!

The biggest problem is IMO the delay (20 to 30seconds). Many people believe the majority of the stream has no idea how it works. Even with democracy implemented, the delay is still a blocker.

But I have no idea how this could be solved (best thing would be multicast directly from source to viewers, but this doesn't work on the open internet, which it why sites like twitch exist).

Using the twitch chat for anything constructive is guaranteed to fail. pls no coperino ravioli.

What does pls no coperino mean? I googled it and see it repeated often (pls no coperino pasterino and other variants).

I'm guessing it means copypasta (copy/paste)

Correct. It came from people copy paste spamming funny comments of others. Its kind of like a joke when you write a long chat message to say it.

My favorite quote:

> If trolls have absolute power, they will use it and they will use it without mercy.

Proof that crowdsourcing doesn't work in every damn situation.

more like proof that running commands over a scrambled channel doesn't work in every damn situation

Since when was this ever a crowd-sourcing effort?

I mean crowd-sourcing in the broadest sense: Making optimal choices based on the wisdom of crowds.

But these aren't optimal choices; they're just the first choices to hit the system.

Guess I misunderstood how it worked. I figured the decision in the game would be an average of the individual player's votes. Something like a democracy. Didn't know it was just whoever pushed the button first.

There's 2 modes. One is anarchy which is exactly how it sounds meaning it takes commands straight from chat. The other mode is democracy which has a simple voting mechanic. The fun is increased because players can vote for the mode they want in the same chat.

But it is actually working, which makes it even more amazing.

should have just had a vote every 10 seconds... most voted answer wins?

Currently, the stream appears to be implementing a hybrid of aggregate vote and random vote (democracy vs. anarchy). Oddly enough, the aggregate vote is doing worse than the random vote.

>Oddly enough, the aggregate vote is doing worse than the random vote

It's simply because aggregate makes the game so much easier that it makes the stream seem less interesting.

Progress is not inevitable with the original system, it's miraculous. With democracy it actually plays the game somewhat normally.

I think it's because of the streaming lag. The stream is several seconds behind the chat and people are submitting commands based off of where the stream is, not where the game is.

A version that played the game on your local PC (with periodic save state syncs maybe) would probably be more robust (and quicker to play).

From the BBC article:

> A five-second rule was introduced at one stage to try to bring some order as several players typed commands at the same time. Whichever command was typed the most over a five-second period was carried out.


Unsure what happened to it, though.

People were generally opposed because it felt like "cheating". I'm sort of on this side -- I think the "anarchy" experiment is pretty fun to watch, though I do think a "democracy" experiment would be interesting to run separately.

The stream can now vote between the two modes and it'll switch if one receives enough votes.

The voting interval made the game too slow to be enjoyable, so players rebelled by spamming the start button until the creator rolled back the change.

If enough players vote for democracy a 5 second rule is implemented.

This is what happens with democracy when you execute it badly.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact