Bitcoin Forum
June 17, 2024, 05:43:05 AM *
News: Voting for pizza day contest
 
   Home   Help Search Login Register More  
Pages: « 1 [2] 3 »  All
  Print  
Author Topic: human mineable and proof of play Altcoins  (Read 3204 times)
HunterMinerCrafter
Sr. Member
****
Offline Offline

Activity: 434
Merit: 250


View Profile
July 14, 2014, 07:44:57 PM
 #21

HunterMinerCrafter, I agree with your explanations:

MOTO is the only true PoP, however I feel there has to be a distinction between HUC and the rest. After all, the game is actually embedded in the blockchain and it can be used to generate coins, even if the game itself doesn't secure the network.

Yes, and I think this aligns with my prior definitions, though I'm not sure what "the rest" means to refer to here.  There are only HUC and MOTO!  Grin

Quote
Maybe we can extend the definition of "mining" to any means of "creating coins", so in this case HUC could be "human mineable" but not "Proof of Play". Or we could make up another term, whatever, but it deserves a distinction.

Yes, in fact I think even with these two new terms we still have not entirely elided the distinctions.  Prior to these third gen coins the "mining" term referred to both coin-base claim semantics and chain-depth security semantics, because they were always the same network rule to accomplish both.  (Some form of either "Find partial collide => increment depth and claim coin" or "be next stake signatory => increment depth and claim coin" generally)  The particular distinction that HUC makes is that it explicitly separates out additional rules for coin claiming, that have nothing to do, directly, with the security threshold function. ("Find partial collide => increment depth and claim coin" with the alternative "Bank at base => claim coin" included.)

Currently, MOTO does still have this symmetry in semantics between coin claim and security function, but it is likely that this will not remain the case for long.

Quote
I'd like to add that I'd bet that all PoP will eventually be mineable by bots, which would eventually outplay humans and make human-mining impossible or at least unprofitable. I like the concept and I think it can be useful to push the boundaries of AI, but don't expect to make money by playing, at least not in the long term.

Again, I argue that any human mine-able coin which is *not* being also mined by bots is inherently insecure, and will operate at the whims of the humans who are most effective at mining it.  When MOTO was new, and bots were still not dominant, skilled players could easily 51% attack the network simply by waiting for "off peak play hours" when competition was naturally low.  Even if there is "round the clock play" by some seasoned players, if there is not continuous and automated competition for those most-skilled players to work against they could simply collude and coordinate their tx selection to control the network.  Basically, any coin which is solely human mined is inherently centralized at the point of the most skilled humans playing at any given moment.

The challenge will be twofold.  In the short term, we must find mechanisms which allow for both bots and humans to be able to mine "peacefully" together.  HUC accomplishes this very well, as a side effect of design.  In MOTO, this is not quite the case and bot operation has significantly interfered with human miners.  I have a proposal to rectify that for MOTO, but the lead dev is still skeptical about it.  (Though he hasn't yet been able to clearly explain why, afaict.)  The longer-term challenge will be in finding game designs that are both fun and "evenly matched" for an AI and a human, so that a bot/human distribution ratio can be arrived at naturally, as in HUC, and not explicitly constrained, as in my proposal for MOTO.  This second part will be a significant challenge!
domob
Legendary
*
Offline Offline

Activity: 1135
Merit: 1166


View Profile WWW
July 15, 2014, 02:18:28 PM
 #22

Again, I argue that any human mine-able coin which is *not* being also mined by bots is inherently insecure, and will operate at the whims of the humans who are most effective at mining it.  When MOTO was new, and bots were still not dominant, skilled players could easily 51% attack the network simply by waiting for "off peak play hours" when competition was naturally low.  Even if there is "round the clock play" by some seasoned players, if there is not continuous and automated competition for those most-skilled players to work against they could simply collude and coordinate their tx selection to control the network.  Basically, any coin which is solely human mined is inherently centralized at the point of the most skilled humans playing at any given moment.

Interesting discussion!  Let me just join in to correct that in the quote above, you probably mean "Proof-of-Play" instead of "human-minable".  HUC is explicitly not insecure due to (lack of) bots, since the security comes from plain old (merge-mined) hardware PoW.  Also, I have to make the same statement as "always" in this context, namely that I personally am still sceptical about the security of PoP in general - especially when there are bots, IMHO the hashing strength depends critically on the used algorithm in the bot logic.  And I think it is much more likely that an order-of-magnitude improvement in the algorithm is found (or even something vastly better, think O(n^2) vs O(n log n) algorithm or something like that), than order-of-magnitude improvements in hardware.  Even the steps from CPU to GPU to FPGA and final ASICs are somewhat predictable and controlled, while one innovation in the bot algorithm / strategy may put someone into the ability to 99.9% attack the network.

Use your Namecoin identity as OpenID: https://nameid.org/
Donations: 1domobKsPZ5cWk2kXssD8p8ES1qffGUCm | NMC: NCdomobcmcmVdxC5yxMitojQ4tvAtv99pY
BM-GtQnWM3vcdorfqpKXsmfHQ4rVYPG5pKS | GPG 0xA7330737
HunterMinerCrafter
Sr. Member
****
Offline Offline

Activity: 434
Merit: 250


View Profile
July 15, 2014, 04:11:29 PM
 #23

Interesting discussion!  Let me just join in to correct that in the quote above, you probably mean "Proof-of-Play" instead of "human-minable".  HUC is explicitly not insecure due to (lack of) bots, since the security comes from plain old (merge-mined) hardware PoW. 

Formally, yes, I'd agree.  The security of the network only falls over directly if the play is securing the transactions.  Arguably, though, if any human mining system is dominated by a small set of miners this still centralizes issuance of the coin, creating a different (social) risk, even though the direct (technical) risk to tx selection is not there.

Quote
Also, I have to make the same statement as "always" in this context, namely that I personally am still sceptical about the security of PoP in general - especially when there are bots, IMHO the hashing strength depends critically on the used algorithm in the bot logic.

Yes, I agree wholeheartedly here.  Just sticking an arbitrary game challenge in as the proof function doesn't generally work, as most games are not sufficient.  You wouldn't believe how many requests I've received to do something like motocoin, but using something like "temple run" or "tetris" games as the proof function....

You really need a challenge function which is "perfect information" and for which the solution finding is sufficiently difficult.  (For now, anyway.  Some more contemporary crypto technologies might change this in the future.)

Quote
  And I think it is much more likely that an order-of-magnitude improvement in the algorithm is found (or even something vastly better, think O(n^2) vs O(n log n) algorithm or something like that), than order-of-magnitude improvements in hardware.

In the case of MOTO, I'd disagree that this is likely.  This was actually my biggest skepticism about MOTO, initially, and I of course set about trying to find such an ideal solver.  "Breaking" the map generation step is isomorphic to breaking sha512, so this is obviously not an option, making the physics simulation itself the only viable target.  The physics simulation is actually fairly strong, with a wide breadth of iterated multipliers creating both a very high cycle requirement and a cascading output effect.  In a sense, the MOTO challenge is actually just a large hierarchy of factorization problems, and I doubt anyone is going to make any sudden strides in factoring.

Of course there is the possibility for better specialized learners, and this is even likely to occur.  However, I don't expect these approaches to be able to suddenly push a MOTO solver into a new complexity family somehow.

If you think you have an idea for an approach that could, I'm all ears!

Quote
  Even the steps from CPU to GPU to FPGA and final ASICs are somewhat predictable and controlled,

I'm not sure that I'd agree, here, but that is an entirely different sort of conversation, and mostly out of context.

Quote
while one innovation in the bot algorithm / strategy may put someone into the ability to 99.9% attack the network.

Only if we are talking about a significant change in expected computational complexity.  If someone could establish a "reverse the physics" solution for MOTO which did not bump into a significant factorization challenge, for example.  I'm highly skeptical that this could happen for the MOTO function, but it is certainly a possibility.  By some argument it is also "possible" that someone finds a trivial collision generation for sha2, scrypt, etc.  (Of course if this were to happen it would spell bigger problems for the world than just the failure of bitcoin and the other crypto currency networks.)

In any case, part of the reasoning behind my recent patch proposals to MOTO is that if any such ideal solver were found, the network would just revert to a classic PoW challenge to keep the network secure until the proof function could be adjusted.  After MOTO does finally fork onto the new network rules an attacker would need to not only find this ideal solver, but then would also need to overcome this additional work requirement, which we can assume is simply not going to happen.  I'd imagine that future proof-of-play coins would also include such "work fallback" safeguards.
gatra
Hero Member
*****
Offline Offline

Activity: 583
Merit: 505


CTO @ Flixxo, Riecoin dev


View Profile WWW
July 15, 2014, 05:32:20 PM
 #24

Yes, I agree wholeheartedly here.  Just sticking an arbitrary game challenge in as the proof function doesn't generally work, as most games are not sufficient.  You wouldn't believe how many requests I've received to do something like motocoin, but using something like "temple run" or "tetris" games as the proof function....

You're right, however with some ingenuity lots of games could be adapted to fit the needs... tetris!? that's quite feasible... I'd do it but it would quickly turn hard to play and become a race on how quickly could bots iterate thru nonces until getting the easiest game... (kinda like MOTO) in this case the challenge to "find mechanisms which allow for both bots and humans to be able to mine "peacefully" together"... would be too hard I'm afraid


would there really be interest in a tetris coin?


           ▄▄▄██████████▄▄▄
       ▄▄██
██████████████████▄▄
     ▄█
█████▀████████████▀██████▄
   ▄█
█████████████████████████████▄
  ▄█
█████████▄█▀▀██████████████████▄
 ▄█
███████████▀██████▄▄█████▄███████▄
▄█
██████████▀██▄▄▄▄██▀▀▀▀▀███████████▄
█████████████▀▀██▀████████▀▀████████
█████████████▄█▀████████████████████
████████▀▀▀▀██▀▀▀▀██████████████████
▀█
██████▀▀▀▀██▀▀▀▀███████████████████▀
 ▀█
███████▄████▄▄███████████████████▀
  ▀█
███████████████████████████████▀
   ▀█
█████████████████████████████▀
     ▀█
█████▄████████████▄██████▀
       ▀▀██
██████████████████▀▀
           ▀▀▀██████████▀▀▀
riecoin       ▄▄█████████▄▄
    ▄██▀▀         ▀▀██▄
  ▄██▀              ▀██▄
 ▄██     ██▄▄          ██▄
▄██      █████▄▄        ██▄
██       ████████▄▄      ██
██       ███████████▄    ██
██       ██████████▀     ██
▀██      ███████▀       ██▀
 ▀██     ████▀         ██▀
  ▀██▄   █▀          ▄██▀
    ▀██▄▄         ▄▄██▀
       ▀▀█████████▀▀
.flixxo   
snailbrain
Legendary
*
Offline Offline

Activity: 1807
Merit: 1020



View Profile
July 15, 2014, 06:29:18 PM
 #25

Asteroid-Coin

Based on the classic game:

https://www.youtube.com/watch?v=OSLZmtwJHJw

--

In motocoin the problem for the "finding maps" is you can fall to the coin - even if you stick in the obstacles i suggested it probably still won't work (although maybe make more difficult if the angle of the fall is much greater as suggested [to find a map]). I think Huntercraftmine or the others will fix this eventually though.

--

If based on asteroid - without being able to fire.. the start can be in the bottom left, the coin can be in the top right..
Asteroids then scroll across the screen (can even scroll both ways - similar to frogger)

https://www.youtube.com/watch?v=l9fO-YuWPSk

The asteroids which scroll across : Many are generated (lots off screen).. so that computing a path is more difficult (time consuming).

There could also be some asteroids which have a "stronger gravity" pull .. e.g. 1 in 6 asteroids. .. and even mini worm holes with very strong gravity. To add some complexity.

Obviously, everything is bot-able  - not sure how long it would last

increasing difficulty could add more "gravity asteroids" and make them scroll faster

HunterMinerCrafter
Sr. Member
****
Offline Offline

Activity: 434
Merit: 250


View Profile
July 15, 2014, 07:20:25 PM
 #26

You're right, however with some ingenuity lots of games could be adapted to fit the needs... tetris!? that's quite feasible... I'd do it but it would quickly turn hard to play and become a race on how quickly could bots iterate thru nonces until getting the easiest game... (kinda like MOTO) in this case the challenge to "find mechanisms which allow for both bots and humans to be able to mine "peacefully" together"... would be too hard I'm afraid

Tetris can't work because there is a deterministic optimal solution if you know the upcoming sequence of blocks - which you must.  The search space for each block state is only a few thousand nodes, so although the complexity of the solver is exponential in the number of pieces used the complexity curve has a low coefficient, it is not steep.  (You'd need some infeasably large game before it would become any sort of challenge at all for a bot, at which point it becomes pointless to humans as a game.)  It wouldn't even be a matter of level iteration to find an easy level, you could just "auto solve" any given level.  A game where the primary challenge is any form of "the player doesn't know what's coming next" simply doesn't work.

Quote
would there really be interest in a tetris coin?

I was asked about it, so at least one random internet user would be interested.

My suspicion is that any game that people tend to pride themselves on being "really awesome at" would have interest.  This probably means most any game.
gatra
Hero Member
*****
Offline Offline

Activity: 583
Merit: 505


CTO @ Flixxo, Riecoin dev


View Profile WWW
July 15, 2014, 07:31:05 PM
Last edit: July 15, 2014, 07:46:55 PM by gatra
 #27


Tetris can't work because there is a deterministic optimal solution if you know the upcoming sequence of blocks - which you must.  The search space for each block state is only a few thousand nodes, so although the complexity of the solver is exponential in the number of pieces used the complexity curve has a low coefficient, it is not steep.  (You'd need some infeasably large game before it would become any sort of challenge at all for a bot, at which point it becomes pointless to humans as a game.)  It wouldn't even be a matter of level iteration to find an easy level, you could just "auto solve" any given level.  A game where the primary challenge is any form of "the player doesn't know what's coming next" simply doesn't work.


but we could adapt the game: show the player all the future pieces and add a "rewind" button as in MOTO... anyway... I'm just saying it might be possible and fun (until bots)... we agree bots would win...
EDIT: I read somewhere that tetris is NP-complete, so yes: the solver is exponential

I was asked about it, so at least one random internet user would be interested.

hehe, not enough for me Smiley but I'd think about it if there's more than that


           ▄▄▄██████████▄▄▄
       ▄▄██
██████████████████▄▄
     ▄█
█████▀████████████▀██████▄
   ▄█
█████████████████████████████▄
  ▄█
█████████▄█▀▀██████████████████▄
 ▄█
███████████▀██████▄▄█████▄███████▄
▄█
██████████▀██▄▄▄▄██▀▀▀▀▀███████████▄
█████████████▀▀██▀████████▀▀████████
█████████████▄█▀████████████████████
████████▀▀▀▀██▀▀▀▀██████████████████
▀█
██████▀▀▀▀██▀▀▀▀███████████████████▀
 ▀█
███████▄████▄▄███████████████████▀
  ▀█
███████████████████████████████▀
   ▀█
█████████████████████████████▀
     ▀█
█████▄████████████▄██████▀
       ▀▀██
██████████████████▀▀
           ▀▀▀██████████▀▀▀
riecoin       ▄▄█████████▄▄
    ▄██▀▀         ▀▀██▄
  ▄██▀              ▀██▄
 ▄██     ██▄▄          ██▄
▄██      █████▄▄        ██▄
██       ████████▄▄      ██
██       ███████████▄    ██
██       ██████████▀     ██
▀██      ███████▀       ██▀
 ▀██     ████▀         ██▀
  ▀██▄   █▀          ▄██▀
    ▀██▄▄         ▄▄██▀
       ▀▀█████████▀▀
.flixxo   
HunterMinerCrafter
Sr. Member
****
Offline Offline

Activity: 434
Merit: 250


View Profile
July 15, 2014, 07:43:23 PM
 #28

If based on asteroid - without being able to fire.. the start can be in the bottom left, the coin can be in the top right..
Asteroids then scroll across the screen (can even scroll both ways - similar to frogger)

https://www.youtube.com/watch?v=l9fO-YuWPSk

It is doable.  Challenges which involve traversal of some dynamic terrain under some constraints of simulated physics seem to "naturally" fit.

Quote
The asteroids which scroll across : Many are generated (lots off screen).. so that computing a path is more difficult (time consuming).

To a bot, there is no such thing as "off screen" so this really just gives some more disadvantage to humans, unfortunately,

Quote
There could also be some asteroids which have a "stronger gravity" pull .. e.g. 1 in 6 asteroids. .. and even mini worm holes with very strong gravity. To add some complexity.

This, on the other hand, is precisely the sort of thing that does to opposite to bots, makes solutions harder to devise.  Most AI algorithms are very good at very consistent/repetitive tasks.  One big reason MOTO was so quickly botted was that it is a single challenge - move toward coin without becoming dead.  Anything that introduces some additional challenge to be considered significantly complicates creating a successful bot.  We call this "modality" referring to the number of modal states that a solver must "potentially be in" while solving.  MOTO has zero modality, so bots can be more or less internally stateless.  They don't need to remember what they just did or "think about" what *sort* of thing they are going to do next because it is always the same one thing - move toward coin without becoming dead.  If a bot has to decide if it is next going to move toward the coin or move away from the coin to avoid an oil slick that will cost it time (just as a wild example) the AI solver will become notably less effective.

Part of why HUC is not just entirely dominated by bots is the fact that the bots must deal with more than just "go pick up coin from spawn and return to base."  They need to decide when to use avoidance tactics to not get blown up by hostile generals.  If someone else gets blown up, they need to decide if it is worth moving in to try to grab the dropped coin.  More advanced bots will manage multiple bots in coordination, with different roles or even complex role states.  (The bot might need to decide when to turn an aggressive hunter general bot into a coin grabber bot, or vice versa.  The bot might need to decide that one bot should stop picking up coins and defend another bot that is already holding many coins, etc.  There are many "potential internal states" to be considered, making bots far less trivial.)

Quote
Obviously, everything is bot-able  - not sure how long it would last

As I've said before, on a long enough curve bots will outperform humans on any given task.  Trying to devise some mechanism for excluding bots entirely is an exercise in futility.

Quote
increasing difficulty could add more "gravity asteroids" and make them scroll faster

As we've now learned from MOTO, it is important to be very very careful with the network's definition of difficulty of proof over block depth.  The more thought I give to these topics the more that I feel that having simple and well defined difficulty scales will be both central and crucial to the success or failure of these kinds of systems.  Arguably, having a poorly defined scale of difficulty is the only actual technical mistake the MOTO developers made, and it created a multitude of problematic side effects that we're now retroactively repairing piece-wise.

HunterMinerCrafter
Sr. Member
****
Offline Offline

Activity: 434
Merit: 250


View Profile
July 15, 2014, 07:50:47 PM
 #29

but we could adapt the game: show the player all the future pieces and add a "rewind" button as in MOTO... anyway... I'm just saying it might be possible and fun (until bots)... we agree bots would win...

You mistook me.  Tetris could only offer "trivial" proofs, and trivial proofs can not secure depth on the chain.

Any game which could be solved "virtually instantly" by even low resource computing devices doesn't offer any security at all.  (Domob's concern is that any given game might eventually be found to be in this category, but I don't necessarily share this concern.  For example, many games are formally reducible to the 3sat problem and as such could be asserted to be sufficiently strong... unfortunately none of the ones I know of offhand would be very fun, they are really just academic exercises.)

Quote
hehe, not enough for me Smiley but I'd think about it if there's more than that

I promise you that tetris really cannot work in any reasonable way as the proof function!
gatra
Hero Member
*****
Offline Offline

Activity: 583
Merit: 505


CTO @ Flixxo, Riecoin dev


View Profile WWW
July 15, 2014, 08:30:43 PM
 #30


many games are formally reducible to the 3sat problem and as such could be asserted to be sufficiently strong... unfortunately none of the ones I know of offhand would be very fun, they are really just academic exercises.)
..
I promise you that tetris really cannot work in any reasonable way as the proof function!

it's the other way around: any computer game can be reduced to the 3sat problem, the interesting and not always possible part is reducing 3sat to the game in question. And it turns out it can be done for tetris. All of the following problems for tetris are np complete:
- Maximizing the number of rows cleared while playing the given piece sequence.
- Maximizing the number of pieces placed before a loss occurs.
- Maximizing the number of simultaneous clearing of four rows.
- Minimizing the height of the highest filled grid square over the course of the sequence.

this means tetris could be used to solve 3sat! you said it yourself: the solver is exponential. The steepness of the curve and the required length for the sequence... well, that remains to be seen.... maybe more complexity can be added by making the playing area wider, or some other small variation to the game that would add more possibilities for each move

we may be getting off-topic with the tetris-specific discussion, but I'm exploring the idea because I feel a tetris-coin would be cool, even if bot-dominated and eventually unplayable by non-augmented humans, still cool


           ▄▄▄██████████▄▄▄
       ▄▄██
██████████████████▄▄
     ▄█
█████▀████████████▀██████▄
   ▄█
█████████████████████████████▄
  ▄█
█████████▄█▀▀██████████████████▄
 ▄█
███████████▀██████▄▄█████▄███████▄
▄█
██████████▀██▄▄▄▄██▀▀▀▀▀███████████▄
█████████████▀▀██▀████████▀▀████████
█████████████▄█▀████████████████████
████████▀▀▀▀██▀▀▀▀██████████████████
▀█
██████▀▀▀▀██▀▀▀▀███████████████████▀
 ▀█
███████▄████▄▄███████████████████▀
  ▀█
███████████████████████████████▀
   ▀█
█████████████████████████████▀
     ▀█
█████▄████████████▄██████▀
       ▀▀██
██████████████████▀▀
           ▀▀▀██████████▀▀▀
riecoin       ▄▄█████████▄▄
    ▄██▀▀         ▀▀██▄
  ▄██▀              ▀██▄
 ▄██     ██▄▄          ██▄
▄██      █████▄▄        ██▄
██       ████████▄▄      ██
██       ███████████▄    ██
██       ██████████▀     ██
▀██      ███████▀       ██▀
 ▀██     ████▀         ██▀
  ▀██▄   █▀          ▄██▀
    ▀██▄▄         ▄▄██▀
       ▀▀█████████▀▀
.flixxo   
HunterMinerCrafter
Sr. Member
****
Offline Offline

Activity: 434
Merit: 250


View Profile
July 15, 2014, 09:12:45 PM
 #31

it's the other way around: any computer game can be reduced to the 3sat problem,

This is not entirely true.  In particular, a game on an arbitrarily large, interactive computing machine with no halting condition (i.e. blockchain) may correspondingly have no finite circuit representation.  This is mostly just minutia though.  You know what I meant.

Tetris may be NP (I haven't seen a formal reduction, but I'd love to if you can dig up a reference) but it is still "too trivial."  For a standard game, the graph is only something like a few thousand nodes, so it is not a very reasonable candidate for the proof function in any case.  You could do a human-mined tetris with a backend PoW like HUC does, but you couldn't do proof-of-play specifically, and mining production would likely be massively overwhelmed by bots.

Quote
this means tetris could be used to solve 3sat! you said it yourself: the solver is exponential. The steepness of the curve and the required length for the sequence... well, that remains to be seen.... maybe more complexity can be added by making the playing area wider, or some other small variation to the game that would add more possibilities for each move

Sure, but again like NP minesweeper the problem itself is so small in the reduction that the resulting curve (relative to modern hardware) is just not nearly steep enough to be useful as a show of expenditure of time.  You'd need a ridiculously large tetris/minesweeper game for it to be useful for proof-of-play at which point you're just throwing human mining out of the window, again.

You could probably modify some game rules (pretty heavily) and make something work out, but it probably wouldn't feel much like tetris in the end.

Quote
we may be getting off-topic with the tetris-specific discussion, but I'm exploring the idea because I feel a tetris-coin would be cool, even if bot-dominated and eventually unplayable by non-augmented humans, still cool

It is actually a good example for discussion, though.  I hope you agree that proof-of-play tetris is right out.  It may be more difficult to see why human-mined tetris doesn't really work out well, but I am pretty sure that there is not much of a reasonable way to balance the bot/human mining.  The bots will just inevitably be able to do the pattern matching much more efficiently.  Convince me otherwise?

gatra
Hero Member
*****
Offline Offline

Activity: 583
Merit: 505


CTO @ Flixxo, Riecoin dev


View Profile WWW
July 15, 2014, 09:55:46 PM
 #32

the tetris paper:
http://erikdemaine.org/papers/Tetris_COCOON2003/paper.pdf

I hope you agree that proof-of-play tetris is right out.

aargh! stop using sentences like that! you are forcing me to make tetriscoin! Smiley

The bots will just inevitably be able to do the pattern matching much more efficiently.  Convince me otherwise?
we agree on that, no need to convince you

apparently, bejeweled and "candy crush saga" fall in the same category: np complete, which means it would work if you use the correct difficulty function, but bots would kill humans anyway

arimaa (http://en.wikipedia.org/wiki/Arimaa) is easy for humans and hard for computers, but the problem is how to transform header hashes to game positions and how make a difficulty function




           ▄▄▄██████████▄▄▄
       ▄▄██
██████████████████▄▄
     ▄█
█████▀████████████▀██████▄
   ▄█
█████████████████████████████▄
  ▄█
█████████▄█▀▀██████████████████▄
 ▄█
███████████▀██████▄▄█████▄███████▄
▄█
██████████▀██▄▄▄▄██▀▀▀▀▀███████████▄
█████████████▀▀██▀████████▀▀████████
█████████████▄█▀████████████████████
████████▀▀▀▀██▀▀▀▀██████████████████
▀█
██████▀▀▀▀██▀▀▀▀███████████████████▀
 ▀█
███████▄████▄▄███████████████████▀
  ▀█
███████████████████████████████▀
   ▀█
█████████████████████████████▀
     ▀█
█████▄████████████▄██████▀
       ▀▀██
██████████████████▀▀
           ▀▀▀██████████▀▀▀
riecoin       ▄▄█████████▄▄
    ▄██▀▀         ▀▀██▄
  ▄██▀              ▀██▄
 ▄██     ██▄▄          ██▄
▄██      █████▄▄        ██▄
██       ████████▄▄      ██
██       ███████████▄    ██
██       ██████████▀     ██
▀██      ███████▀       ██▀
 ▀██     ████▀         ██▀
  ▀██▄   █▀          ▄██▀
    ▀██▄▄         ▄▄██▀
       ▀▀█████████▀▀
.flixxo   
snailbrain
Legendary
*
Offline Offline

Activity: 1807
Merit: 1020



View Profile
July 15, 2014, 10:52:03 PM
Last edit: July 16, 2014, 07:35:45 AM by snailbrain
 #33

If based on asteroid - without being able to fire.. the start can be in the bottom left, the coin can be in the top right..
Asteroids then scroll across the screen (can even scroll both ways - similar to frogger)

https://www.youtube.com/watch?v=l9fO-YuWPSk

It is doable.  Challenges which involve traversal of some dynamic terrain under some constraints of simulated physics seem to "naturally" fit.

Quote
The asteroids which scroll across : Many are generated (lots off screen).. so that computing a path is more difficult (time consuming).

To a bot, there is no such thing as "off screen" so this really just gives some more disadvantage to humans, unfortunately,

Probably misunderstood (or maybe you didn't and you're right) - the asteroids are in rows (scrolling): if there are 1000 asteroids in a row  as opposed to 10 asteroids in a row (10 on the screen at all times, all of which have random gravities or effects (maybe some have a pulsating gravity?), it may be more difficult to traverse due to having to calculate more objects (the rows would be moving different speeds like frogger).
But thinking more -- maybe there is more chance of easy gaps... and/or maybe it doesn't matter how many there are if it calculates in real-time objects which are near..

Quote
arimaa (http://en.wikipedia.org/wiki/Arimaa (http://en.wikipedia.org/wiki/Arimaa)) is easy for humans and hard for computers, but the problem is how to transform header hashes to game positions and how make a difficulty function

still a prize - till 2020
http://arimaa.com/arimaa/challenge/2014/

Quote
increasing difficulty could add more "gravity asteroids" and make them scroll faster

just thought - scrolling faster would do nothing (unless each row scrolled different speeds?)- more gravity stuff may -- also a ufo which fires from a middle row Cheesy


HunterMinerCrafter
Sr. Member
****
Offline Offline

Activity: 434
Merit: 250


View Profile
July 15, 2014, 10:53:05 PM
 #34

the tetris paper:
http://erikdemaine.org/papers/Tetris_COCOON2003/paper.pdf

I hope you agree that proof-of-play tetris is right out.

aargh! stop using sentences like that! you are forcing me to make tetriscoin! Smiley

Go for it, I'm just saying that it would have to be just a PoW coin and not really a human-mined/PoP coin.  Nobody in their right might would ever "play" that large-enough-to-be-secure tetris game in the same way that nobody in their right mind would calculate a sha hash collision manually.

Quote
The bots will just inevitably be able to do the pattern matching much more efficiently.  Convince me otherwise?
we agree on that, no need to convince you

Dang, I was actually hoping you could. Wink

Quote
apparently, bejeweled and "candy crush saga" fall in the same category: np complete, which means it would work if you use the correct difficulty function, but bots would kill humans anyway
Particularly because again much of the fun of the challenge to a human is in not knowing what is coming up next.

Quote
arimaa (http://en.wikipedia.org/wiki/Arimaa) is easy for humans and hard for computers, but the problem is how to transform header hashes to game positions and how make a difficulty function

Yes, this has lately become the canonical example of "hard for bot" game.  I've been toying with the idea of "whipping this up" as a fork of MOTO, just to illustrate the point.  Both of the problems you mention are pretty easily met with basic solutions.  It wouldn't be great, but it would work.

The problem would be in taking the security of such a coin seriously.  Although we believe Arimaa to be difficult, I'd actually have an easier time believing it could be "suddenly solved" in the way Domob might suggest than believing MOTO could be. (I'm probably wrong, but we can't really confidently say which is "stronger" and therein lies the rub.  Wink)

gatra
Hero Member
*****
Offline Offline

Activity: 583
Merit: 505


CTO @ Flixxo, Riecoin dev


View Profile WWW
July 16, 2014, 02:43:01 AM
 #35

Quote
arimaa (http://en.wikipedia.org/wiki/Arimaa) is easy for humans and hard for computers, but the problem is how to transform header hashes to game positions and how make a difficulty function

Yes, this has lately become the canonical example of "hard for bot" game.  I've been toying with the idea of "whipping this up" as a fork of MOTO, just to illustrate the point.  Both of the problems you mention are pretty easily met with basic solutions.  It wouldn't be great, but it would work.


I'm intrigued... I can think of many ways of transforming hashes to game positions, but I don't imagine how would one decide if the work done is enough for declaring a block.. some positions will be trivial loses, some trivial wins, and some very intricate. Would the work be maximizing some fixed "score" function?

Quote
The problem would be in taking the security of such a coin seriously.  Although we believe Arimaa to be difficult, I'd actually have an easier time believing it could be "suddenly solved" in the way Domob might suggest than believing MOTO could be. (I'm probably wrong, but we can't really confidently say which is "stronger" and therein lies the rub.  Wink)

of course it depends on how is arimaa implemented, however I'd say it's more likely to have a sudden breakthru in MOTO: some general strategy like "stay balanced, don't go too fast, and aim for the coin". Arimaa looks like chess: it's just exponential and you can have heurisitics but at the end of the day you have to explore a lot of positions.


           ▄▄▄██████████▄▄▄
       ▄▄██
██████████████████▄▄
     ▄█
█████▀████████████▀██████▄
   ▄█
█████████████████████████████▄
  ▄█
█████████▄█▀▀██████████████████▄
 ▄█
███████████▀██████▄▄█████▄███████▄
▄█
██████████▀██▄▄▄▄██▀▀▀▀▀███████████▄
█████████████▀▀██▀████████▀▀████████
█████████████▄█▀████████████████████
████████▀▀▀▀██▀▀▀▀██████████████████
▀█
██████▀▀▀▀██▀▀▀▀███████████████████▀
 ▀█
███████▄████▄▄███████████████████▀
  ▀█
███████████████████████████████▀
   ▀█
█████████████████████████████▀
     ▀█
█████▄████████████▄██████▀
       ▀▀██
██████████████████▀▀
           ▀▀▀██████████▀▀▀
riecoin       ▄▄█████████▄▄
    ▄██▀▀         ▀▀██▄
  ▄██▀              ▀██▄
 ▄██     ██▄▄          ██▄
▄██      █████▄▄        ██▄
██       ████████▄▄      ██
██       ███████████▄    ██
██       ██████████▀     ██
▀██      ███████▀       ██▀
 ▀██     ████▀         ██▀
  ▀██▄   █▀          ▄██▀
    ▀██▄▄         ▄▄██▀
       ▀▀█████████▀▀
.flixxo   
HunterMinerCrafter
Sr. Member
****
Offline Offline

Activity: 434
Merit: 250


View Profile
July 16, 2014, 04:44:50 AM
 #36

I'm intrigued... I can think of many ways of transforming hashes to game positions, but I don't imagine how would one decide if the work done is enough for declaring a block.. some positions will be trivial loses, some trivial wins, and some very intricate.

The easy solution is to just assume all boards comparable on average and require some number of consecutively generated boards be solved for in a row instead.

Quote
Would the work be maximizing some fixed "score" function?

This would be the less easy solution, requiring solutions to meet some threshold of "quality."  This is complicated by the relative weighting of significance of Arimaa pieces being generally unknown, so scoring would always have to be assumed to have a margin of error.

Quote
of course it depends on how is arimaa implemented, however I'd say it's more likely to have a sudden breakthru in MOTO: some general strategy like "stay balanced, don't go too fast, and aim for the coin".

This is pretty much the approach taken now, but it is one of those things easier said than done.  The constraints of the problem really do complicate things nicely.  The noisy terrain and limited movement combine to create (more commonly than you'd probably think) situations where a map looks perfectly reasonable but is actually almost or entirely non-solvable, simply because there isn't a path that fits within the physics constraints.  Because the search space on paths is so enormously huge and the tiniest detail of the perlin map can cascade into a drastically different resultant path with even just one frame of a difference in input timing I don't expect much in the way of a total/"breaking" solver to be devised.  We will get progressively better learners, but I would be very surprised if someone showed up with a useful deterministic (or even hybrid) solver.  General search optimization (annealing) is probably even the best anyone *can* do, to do otherwise would likely require devising some way "around" the factorization problem.  (Granted, people could be doing their optimizing searches a lot more efficiently than anyone seems to be now.)

I'd encourage you to try to devise a solver, however, it is a lot of fun to work through!  It is also quite fascinating to see "hands on" how the integration combines with the perlin function to create the security offered by the proof.  (That "A HA moment" when you realize how, as a hash function, it makes even these crazy chained X-whatever hashing schemes look downright weak.....)

Quote
Arimaa looks like chess: it's just exponential and you can have heurisitics but at the end of the day you have to explore a lot of positions.

The same can be said for MOTO, both games are highly branching right from the initial states.  If you haven't tried it in ForFun mode (just run the motogame binary not through the wallet) you should give it a shot.  Try to get to the coin with 40-50 seconds left on the clock (as on the live network) instead of just getting to the coin at all.  It is much more of a "puzzle" game than it might seem.  Wink
HYPERfuture
Hero Member
*****
Offline Offline

Activity: 686
Merit: 500

HYPER project manager and PR + GoldPieces [GP]


View Profile WWW
July 16, 2014, 05:53:26 AM
 #37

There are a few misonceptions about HYPER that have been raised in this thread I would like to clear up.

The biggest one being the funds HYPER has set aside for MMO Development, bounties and to sponsor more games and game development with the HYPER incubator funds.

There are less than 700 000 HYPER remaining in the various funds (this is entirely transparent and can be verified via the blockchain). Of this 700 000, the 500 000 MMO Development Fund is being escrowed and managed by psybits, a respectable Hero Member on the forum.

Of the remaining 200 000, it is being used to sponsor more game developments (such as the HYPER Counter Strike. Minecraft and Zandagort servers), and also for bounties anyone can participate in such as these: https://bitcointalk.org/index.php?topic=695339.0 So there are no millions of coins to be dumped, management of the funds has been entirely transparent and can be seen at any time. Please try reading the HYPER thread or OP from time to time if you are going to judge our coin.

Through sponsoring as many online games and servers as possible, HYPER aims to be a cryptocurency that is not just for speculation, but is embedded in a complex ecosystem with many opportunities for playing, trading, staking, gaming, and earning!

Biggest current HYPER developments:

- New website design coming out soon

- HYPER Reddit tip bot was just launched

- Testing to add HYPER to the in-game free market trading platform in space strategy MMO Zandagort we are running  is underway

- Counter Strike tournaments sponsored by HYPER are starting very soon

- A monthly HYPER Counter Strike Blitz that rewards the top players on the CS servers is also launching very soon

- I am finishing up the HYPER Whitepaper that will discuss in more detail the different HYPER funds. I am also taking feedbck on the HYPER thread regarding how the funds should be managed.

- There are many more partnerships with online games on the way that I cannot announce yet as we are still working out the details.

Thanks for reading guys Smiley What makes HYPER unique is the funds we have at our disposal - so the number of online games that incorporate HYPER and the number of projects we will be sponsoring can only grow - meaning HYPER will soon have one of the most extensive ecosystems of any alt out there!


HYPER Gaming Currency -> https://bitcointalk.org/index.php?topic=624651 GP RPG Currency -> https://bitcointalk.org/index.php?topic=1053441 https://cryptogalaxies.com -> Blockchain Based Space Strategy MMO. Crypto Galaxies on Bitcointalk -> https://bitcointalk.org/index.php?topic=1374470
HunterMinerCrafter
Sr. Member
****
Offline Offline

Activity: 434
Merit: 250


View Profile
July 16, 2014, 06:51:07 AM
 #38

There are less than 700 000 HYPER remaining in the various funds (this is entirely transparent and can be verified via the blockchain).

How can we know what happened to the rest, though?  Presumably some was given away in your games but we have know way to know if you didn't just keep it.

Quote
Of the remaining 200 000, it is being used to sponsor more game developments (such as the HYPER Counter Strike. Minecraft and Zandagort servers),

You're promoting the very model that we're seeking to obviate in this thread, as if to advertise it to us?

This has to be the most misguided thread hijacking that I've seen in awhile.

Quote
and also for bounties anyone can participate in such as these: https://bitcointalk.org/index.php?topic=695339.0 So there are no millions of coins to be dumped, management of the funds has been entirely transparent and can be seen at any time. Please try reading the HYPER thread or OP from time to time if you are going to judge our coin.

Personally, I've watched the hyper thread since you launched.  Mostly out of interest in observing games-based issuance being done in precisely the wrong way.  I would hardly say there is any transparency there into the redistribution of your 1.5 millions of coins.

Quote
Through sponsoring as many online games and servers as possible, HYPER aims to be a cryptocurency that is not just for speculation, but is embedded in a complex ecosystem with many opportunities for playing, trading, staking, gaming, and earning!

All without any consideration for the original design goals of crypto currencies - fair and transparent issuance, total decentralization of transaction, and open accountability of history.  Your "ecosystem" even directly renegs on all three!

Quote
Biggest current HYPER developments:

Are you doing anything innovative with crypto tech?

Quote
Thanks for reading guys Smiley What makes HYPER unique is the funds we have at our disposal - so the number of online games that incorporate HYPER and the number of projects we will be sponsoring can only grow - meaning HYPER will soon have one of the most extensive ecosystems of any alt out there!

Now your premine is somehow "unique" too?  Hardly.  I used to at least get a laugh or two out of all of the HYPER hype.  Lately I can't even chuckle at it anymore, it is just sad.

Good luck with your coin and your central servers and your whatever PR machine tactics you'll be using tomorrow.  I hope the model works out for you, somehow or another.

We'll be over here working on something just a little bit different in the meantime.
domob
Legendary
*
Offline Offline

Activity: 1135
Merit: 1166


View Profile WWW
July 16, 2014, 07:03:37 AM
 #39

The physics simulation is actually fairly strong, with a wide breadth of iterated multipliers creating both a very high cycle requirement and a cascading output effect.  In a sense, the MOTO challenge is actually just a large hierarchy of factorization problems, and I doubt anyone is going to make any sudden strides in factoring.

Please don't take my scepticism too "literally" - you've worked on a bot, not me.  And I'm a mathematician, so I tend to "over-abstract" things a lot of times from the point of view of engineers....  I have no doubt that MOTO can be really hard to solve, I believe you about that.  My main point is just that at least in theory, PoW like SHA collisions are more secure since the entire design goal of SHA was to make it hard to break.  With MOTO or any other game, while the game may be hard to bot, it is mostly or at least partly designed to be a fun game.  So inherently it "can't" be as secure as something that's designed only for the very purpose of being hard to break.  (Which is not to say that it is impossible to find a flaw in SHA before finding an efficient MOTO solver - nobody knows for now.  But I deem it less likely.)

I really enjoy the discussions going on here, and I think that the ideas behind both HUC and MOTO are really innovative and may lead to "something" in the future.  It is a perfect fit, IMHO, to apply blockchain technology to replace central servers in multiplayer games.

Use your Namecoin identity as OpenID: https://nameid.org/
Donations: 1domobKsPZ5cWk2kXssD8p8ES1qffGUCm | NMC: NCdomobcmcmVdxC5yxMitojQ4tvAtv99pY
BM-GtQnWM3vcdorfqpKXsmfHQ4rVYPG5pKS | GPG 0xA7330737
cwb27
Full Member
***
Offline Offline

Activity: 148
Merit: 100


View Profile
July 16, 2014, 07:50:56 AM
 #40

so we get a shovel and go to location and mine the coin? human mineable?
Pages: « 1 [2] 3 »  All
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!