# Are we bound to loose?

Thread in 'Discussion' started by Stan64, 9 Apr 2006.

2. ### Stan64Unregistered

There is a Bastet mode in a home version of Tetris that chooses the worst tetrmino for you all the time. Try it out. =O

http://abrick.sourceforge.net/

3. ### colour_thief

Excellent, I've been meaning to talk about this for a while now. Basically: They are wrong and we are not doomed to die. Why?

Well, Tetris games seem nice and random, however they still use a pseudo random number generation. How these typically work is: an extremely large number is generated (a "seed"), and then that number is "hashed" for subsequent pieces. I'm skipping a lot of technical details here, but this hashing will, after many many hashings, bring you back to the original seed. What does this mean? If you played enough, the game would start looping the pieces it gave you.

This is a REALLY big loop, but still a loop. Assuming a 32-bit seed, you're looking at a game looping its pieces after at most ~4.29 billion pieces. That far too absurd to consider it actually happening in a real game, but it has real implications on the claim that "we are doomed to die because we will encounter nothing but S and Z pieces... eventually". There is no eventuality. The game is looped, and if that sequence isn't in the loop, there's no real reason you can't continue playing.

The probability of getting an S or Z piece is 2/7. The probability of getting n such pieces in a row is (2/7)^n. The survival record is what... 229 SZ pieces? And the proof gives an upper bound on survival of about 70 000 pieces? Let's say even 50 would kill you in a normal game. The probability of getting either S or Z 50 times in a row is 6.26 x 10^-28.

So, it's not going to happen. Have fun playing Tetris forever.

4. ### Stan64Unregistered

Yesh. But there still is a chance though?

(I got 5570 in Bastet mode after a couple of tries =)

EDIT: And the site only states a worst case scenario. What if all the bricks wont fit anywhere and fill your screen? It don't have to be a S or Z.

5. ### colour_thief

It's not that there's a chance. If you know the hashing function, it's trivial to prove that it won't happen. I'd be willing to bet it's impossible in every commercial Tetris to date.

I know what bastet is though I've never played it. I bet if you had piece previews the game would be much easier, possibly lasting indefinitely.

6. ### colour_thief

Heh, looking at bastet, I bet that if you set it to always give you the worst piece you could create a pattern that would fool the AI into giving you pieces pieces that would sustain that repeated pattern.

7. ### Stan64Unregistered

Yeah, but my friend implies that what if you get a bad piece then start to fix it but to soon you get another bad piece. Myself thinks that that wouldn't happen with preview and all though.

And in Bastet there is a chance that you get the almost worst piece. So a pattern would not last to long. I thought about that too.

8. ### Needle

The study in question would probably ring true if the randomizer was truly random, instead of pseudo-random like most computer implementations of randomness... For instance, if the randomizer used real-world random stuff like atmospheric noise or radioactive decay, the sequence would be genuinely random and non-looping, eventually ending up in that overwhelming stream of Zs or Ss.

Not that that's going to possibly ever happen before the universe ends, but the possibility would then be non-zero, however incredibly low it may be.

http://random.org/essay.html

9. ### colour_thief

I've heard some games use player input as a source of entropy, though Ive never personally seen one do it.