[ANN][HUC] Huntercoin - Worlds First Decentralized Game/World on the Blockchain

[The Bitcoin Co-op]

.....  and let people play pacman, rogue, whatever.

this would allow it, but no one seems too interested -
http://forum.huntercoin.org/index.php/topic,22249.0.html

I'm interested. I just can't upgrade the Huntercoin client to do that, myself....

I still think the way to do it is ...

4) An important part of the game is that it should not have any aspect of 'waiting for the next block'. One of the biggest annoyances of one of the people I got to play the game was that waiting for a move is not good. Maybe a board starts with a small amount of "potential HUC" and each block updates the "potential HUC" that a player might earn, but the actual gameplay is seamless and does not involve waiting.

This (I believe) would allow bots to easily give themselves a win (aka cheat) every time, it might even be more profitable than playing the original game. I can't think of a situation where this isn't possible.

To be fair I haven't played this game for like, a year so I don't know much about it right now.

You mean because bots can make thousands of moves per second? Yeah, I can see how they might make a lot of coins that way, playing Pac Man a hundred times a second.

On the other hand, everybody loves a good bounty. I'm sure a solution exists. I'm more interested in using mini-games for the purpose of resolving combat between hunters, though... like, it could zoom to a rock-paper-scissors match when they get in fighting range, instead of using Destruct

[domob]

This sounds a lot like what Motocoin tried - which is a very interesting approach, but seemed to not really work out in the end (due to bots).

[jwinterm]

...
I'm sure a solution exists.
...

I don't think this is the case. In order to eliminate bots for a game-based blockchain you need to design a game that meets two criteria:

1) Bots can't "win" the game, or if they can earn some coins, it has to be many orders of magnitude less than a human playing the same game, otherwise someone can just run thousands of bots, right? You basically need a game version of the Turing test.

2) You need to have a game that, although bots suck at it, the results must be able to be easily and quickly verified by a computer. This is another wrinkle that makes point #1 much harder to achieve.

Afaik, a solution does not exist, and if you came up with one you'd probably be some kind of laureate or at least have a cushy CS professorship at MIT or something.

[cryptospreader]

...
I'm sure a solution exists.
...

I don't think this is the case. In order to eliminate bots for a game-based blockchain you need to design a game that meets two criteria:

1) Bots can't "win" the game, or if they can earn some coins, it has to be many orders of magnitude less than a human playing the same game, otherwise someone can just run thousands of bots, right? You basically need a game version of the Turing test.

2) You need to have a game that, although bots suck at it, the results must be able to be easily and quickly verified by a computer. This is another wrinkle that makes point #1 much harder to achieve.

Afaik, a solution does not exist, and if you came up with one you'd probably be some kind of laureate or at least have a cushy CS professorship at MIT or something.

I can suggest these devs can combine a joint huge bounty to hunt for open source solution to feed these concerns. A lot of people will compete.

[wiggi]

That is not quite correct - the client is not graphical at all (also not Windows-based if you mean that) and does not use Qt.  It is just like the Bitcoin Core daemon client (bitcoind) and like the old huntercoind.

That's actually already present - Huntercoin Core includes the Qt UI of Bitcoin Core, which is fully functional (same in Namecoin).  So you can use it to manage your coins (without playing) from there, or use the console window - this may make it straight-forward for you to rebase your gaming UI onto it.

to clean up the confusion, on a new Linux (tested with Mint18 in a VM) the build process for Huntercore is exactly the same as for Bitcoin. Installed dependencies decide of whether the Qt-client is made:

Code:

sudo apt-get install libqt5gui5 libqt5core5a libqt5dbus5 qttools5-dev qttools5-dev-tools libprotobuf-dev protobuf-compiler
./autogen.sh
./configure --with-gui=qt5 --with-incompatible-bdb
make

Need right-click on desktop|Create Launcher to execute the resulting /home/workgroup/huntercore-master/src/qt/huntercoin-qt
(default working dir is .huntercoin but should probably be .huntercore)

To work with the pruned blockchain from http://forum.huntercoin.org/index.php/topic,22255.msg26622.html#msg26622
it needs a huntercoin.conf to get connections and to not insist on reindexing which would take almost as long as chain download, e.g.

Code:

prune=550
addnode=192.168.178.22

It really exists. Huntercore-Qt (with qt5, capable of autodetecting and using all kinds of GPUs)

Awesome

[jwinterm]

...
I'm sure a solution exists.
...

I don't think this is the case. In order to eliminate bots for a game-based blockchain you need to design a game that meets two criteria:

1) Bots can't "win" the game, or if they can earn some coins, it has to be many orders of magnitude less than a human playing the same game, otherwise someone can just run thousands of bots, right? You basically need a game version of the Turing test.

2) You need to have a game that, although bots suck at it, the results must be able to be easily and quickly verified by a computer. This is another wrinkle that makes point #1 much harder to achieve.

Afaik, a solution does not exist, and if you came up with one you'd probably be some kind of laureate or at least have a cushy CS professorship at MIT or something.

I can suggest these devs can combine a joint huge bounty to hunt for open source solution to feed these concerns. A lot of people will compete.

+1

But how would the development be structured if there is just one client, etc? It sounds sort of like the development of capchas, where there is a constant ongoing war between people paid to enforce the integrity of the "human factor", and those who profit from automating a process to outsmart it. In this context, it needs a 'game', or some variable in the game, that humans can learn faster than bots, right?

One of the biggest problems with developing this kind of thing is the motivation to equate "low income workers" with bots.

A coin has the potential to accomplish the rare feat of "profiting" by outsourcing its earning potential to poor countries, but that is perceived as harmful to the coin, when it is actually not. If, for example, HUC were able to 'outsource' all of its human mining to people in poor countries then of course nobody in rich countries would 'mine' HUC, but the algorithm would be producing more "info", more "product" than coins that maintained the exclusivity of their mining to wealthier miners. Productivity of the algorithm, market cap, would sky rocket but there would be no 'human miners' except in poor countries. 

It's not exactly the same as captchas, because for a captcha there is a human setting the correct answer. This isn't really feasible for HUC, because if there was a human setting the correct answer in the code, anyway can look at the code and see what the correct answer(s) is(are) and program their bot(s) accordingly. I think it's really an intractable problem to create something that humans are far better than bots at, can be quickly verified by a computer, and the entire thing contained in open source code that anyone can check.

[wiggi]

...
I'm sure a solution exists.
...

I don't think this is the case. In order to eliminate bots for a game-based blockchain you need to design a game that meets two criteria:

1) Bots can't "win" the game, or if they can earn some coins, it has to be many orders of magnitude less than a human playing the same game, otherwise someone can just run thousands of bots, right? You basically need a game version of the Turing test.

2) You need to have a game that, although bots suck at it, the results must be able to be easily and quickly verified by a computer. This is another wrinkle that makes point #1 much harder to achieve.

Afaik, a solution does not exist, and if you came up with one you'd probably be some kind of laureate or at least have a cushy CS professorship at MIT or something.

I'm also sure a solution exists. (One that could be of actual use for Huntercoin)


(*) my assumptions, didn't play or do source diving.

Motocoin uses block hash (*) to seed the RNG for creating a level, and when the first player has solved it, they submit the result: the complete player input,
keystrokes (*) with timestamps (*) measured in terms of ticks of the game's physics engine (*). Verifying a block means replaying the level to see if the submitted player input solves the level.


The standard example of a un-bottable game would be Nethack, but it takes at least some hours + expert skills to solve.
(result is also a bunch of keystrokes, quickly verified by a computer)

If the game is simpler (but still a turn based "roguelike" game) and bots can win, they need less time than a human, but more game turns.
Example: Angband bot, if very lucky: 10.000.000 turns, Human expert: 50.000 turns

If the game is too short (same turn based "roguelike" game) then bots can simply try every possible action and brute force (if it's possible to win in ~100 turns)
Example: You start as high level char in the deepest Angband dungeon level, but stripped of every possession: try to escape as fast as possible.


Huntercoin doesn't need the result for the next block, it could create a random "instanced area" once per day, hunters that would enter this area would be alone, playing a (local) single player game against (deterministic) computer controlled monsters. Only after winning, results get posted to the blockchain, and only after the day is over, the fastest winner (in terms of game turns) get coins. (game turns != Huntercoin blocks)

The number of game turns would have a (not too low) target value, fast win today == extra nasty dragons tomorrow.

[jwinterm]

...
The standard example of a un-bottable game would be Nethack, but it takes at least some hours + expert skills to solve.
(result is also a bunch of keystrokes, quickly verified by a computer)

If the game is simpler (but still a turn based "roguelike" game) and bots can win, they need less time than a human, but more game turns.
Example: Angband bot, if very lucky: 10.000.000 turns, Human expert: 50.000 turns
...

How can you have a roguelike game on the blockchain and not pass the entire gamestate to the client(s)? And if the client knows the entire gamestate, isn't it trivial to construct a bot that wins rapidly? If you want to design an interactive faucet, ok, roguelikes that only reveal a small portion of the gamestate are fine (I'm actually working on one myself for myriadcoin here http://dungeon.myrcraft.com/ using sockets.io to only transmit the part of the gamestate that the player can "see"), but to have a truly decentralized game, how can you hide portions of the gamestate that would make it unbottable?

[The Bitcoin Co-op]

When I stated that a solution probably exists, I was thinking very long term. In the future, blockchains will all be more advanced as well as more integrated with one another. We could do things like have AI designed to detect bots, and run them off-chain in partnership with some other decentralized platform if that proved intractable to do directly. I think there are also blockchains designed to obfuscate the data they contain--for example, MaidSafe/DECENT/Storj do that just for protocol security/data integrity purposes--such that a captcha system could be feasible.

For now, though, I think we need to stick with gameplay that cannot easily be gamed by bots. A move timer accomplishes that pretty effectively. I think there are other ways in the context of multiplayer games, as well--a human opponent forces the bot to play at normal speed, instead of winning the game every millisecond and raking up coins.

[wiggi]

...
The standard example of a un-bottable game would be Nethack, but it takes at least some hours + expert skills to solve.
(result is also a bunch of keystrokes, quickly verified by a computer)

If the game is simpler (but still a turn based "roguelike" game) and bots can win, they need less time than a human, but more game turns.
Example: Angband bot, if very lucky: 10.000.000 turns, Human expert: 50.000 turns
...

How can you have a roguelike game on the blockchain and not pass the entire gamestate to the client(s)? And if the client knows the entire gamestate, isn't it trivial to construct a bot that wins rapidly? If you want to design an interactive faucet, ok, roguelikes that only reveal a small portion of the gamestate are fine (I'm actually working on one myself for myriadcoin here http://dungeon.myrcraft.com/ using sockets.io to only transmit the part of the gamestate that the player can "see"), but to have a truly decentralized game, how can you hide portions of the gamestate that would make it unbottable?

In a decentralized game, you can't hide portions of the gamestate. If the game is designed to be decentralized, the gamestate can be made to contain as little information as possible (enemies spawn only at FoV perimeter and despawn if not visible for X turns, items in containers spawn only if opened etc)

If the game is off the shelf with only minimal modifications, then players would have clairvoyance from the start and know everything the gamestate knows. I don't think this makes the gap between human and bot performance smaller. The more you know, the more you can make strategic plans that would never occur to a bot. You "see" if a specific level or area is worth cleaning out or not when every turn counts.


Otoh, there has never been a monetary incentive for making such a bot, and I've not yet seen a bot that would do systematic savescumming every time it thinks the RNG made a decision that is less than optimal. No one knows how good these buggers get if coin can be won.


Btw, to clarify: "gamestate" here has nothing to do with the Huntercoin gamestate, right? Only the collected player input (2000 or so keystrokes) would be part of it, until it gets replayed by an encapsulated part of the Huntercoin client (which may mostly consist of the original off the shelf game code). The result of the replaying (just "ok" or "invalid") is also saved in Huntercoin gamestate until reward is paid to the winner with the shortest string of keystrokes.


edit: quickly verified by a computer == CPU budget would be a few microseconds per turn, or <100 milliseconds per game replay, and the code must be trusted to not crash

[Icon]

Thought i would chime in, has anyone considered adding or changing Huntercoin from POW to POS or both that way us bag holders could make some huc while we wait hours to travel the map... Also would not have to depend on 1 pool or a few smaller pools as our workhorses to move the blocks down the line.


Icon

[Icon]

Has anyone got any full nodes up and running i got the list of addnodes from OP on first page and still only showing 2 connections 

Icon

addnode=178.32.31.41
addnode=178.62.17.234
addnode=192.81.209.210
addnode=192.99.247.234
addnode=192.241.222.65

[jwinterm]

...
In a decentralized game, you can't hide portions of the gamestate. If the game is designed to be decentralized, the gamestate can be made to contain as little information as possible (enemies spawn only at FoV perimeter and despawn if not visible for X turns, items in containers spawn only if opened etc)

If the game is off the shelf with only minimal modifications, then players would have clairvoyance from the start and know everything the gamestate knows. I don't think this makes the gap between human and bot performance smaller. The more you know, the more you can make strategic plans that would never occur to a bot. You "see" if a specific level or area is worth cleaning out or not when every turn counts.


Otoh, there has never been a monetary incentive for making such a bot, and I've not yet seen a bot that would do systematic savescumming every time it thinks the RNG made a decision that is less than optimal. No one knows how good these buggers get if coin can be won.
...

That is a good point re: only spawn enemies or items in FoV. Ultimately though, if significant money is at stake, I think bots will be developed that outperform humans.

[Icon]

Thanks! the OP should be update the addnodes went from 2 connections to over 9

Icon

[Vlad2Vlad]

Snailbrain, et al,

How many total individual gamers do you guys have?  How many are active? 

Thx!

[The Bitcoin Co-op]

...
I'm sure a solution exists.
...

I don't think this is the case. In order to eliminate bots for a game-based blockchain you need to design a game that meets two criteria:

1) Bots can't "win" the game, or if they can earn some coins, it has to be many orders of magnitude less than a human playing the same game, otherwise someone can just run thousands of bots, right? You basically need a game version of the Turing test.

2) You need to have a game that, although bots suck at it, the results must be able to be easily and quickly verified by a computer. This is another wrinkle that makes point #1 much harder to achieve.

Afaik, a solution does not exist, and if you came up with one you'd probably be some kind of laureate or at least have a cushy CS professorship at MIT or something.

I can suggest these devs can combine a joint huge bounty to hunt for open source solution to feed these concerns. A lot of people will compete.

Maybe it would be easier for btc co op to handle a bounty since he's done bounties in the past and is widely known.

I actually prefer for SnailBrain to handle bounties. He's the founder, and also, I always find it too tempting to trade with the coins. I'm less wealthy than you think I am. 

[Vlad2Vlad]

I actually prefer for SnailBrain to handle bounties. He's the founder, and also, I always find it too tempting to trade with the coins. I'm less wealthy than you think I am. 

I thought the namecoin founder also founded this coin.  And he's dead. 

Is that not the case?  I don't wanna buy some living guy's coin. 

[jwinterm]

...

I actually prefer for SnailBrain to handle bounties. He's the founder, and also, I always find it too tempting to trade with the coins. I'm less wealthy than you think I am.  

Didn't the founder actually die? Not trying to be pedantic, but I would go with maintainer or lead developer or something.

Edit: http://huntercoin.org/information/mikhail-sindeyev/

[The Bitcoin Co-op]

...

I actually prefer for SnailBrain to handle bounties. He's the founder, and also, I always find it too tempting to trade with the coins. I'm less wealthy than you think I am.  

Didn't the founder actually die? Not trying to be pedantic, but I would go with maintainer or lead developer or something.

Edit: http://huntercoin.org/information/mikhail-sindeyev/

I think SnailBrain actually came up with the concept, but Mikhail was extremely influential and productive, as he was with Namecoin. That's why Namecoin development is stalled. Truly a great loss for everyone

If you're worried about buying a living guy's coin, though, I wouldn't be. What small premine HUC had was also lost with Mikhail's passing, as he also was managing that. This project is therefore as open to you as it is to any of us. I doubt either he or I are the biggest HUC holders, right now.

[The Bitcoin Co-op]

...

I actually prefer for SnailBrain to handle bounties. He's the founder, and also, I always find it too tempting to trade with the coins. I'm less wealthy than you think I am.  

edit
 Whoever holds the bounty is fine.  I'll chip in .02 btc and 2000 HUC and as long as any work done is open source, available to other devs or other projects, the bounty could develop some serious interest as cryptospreader said.

...
I'm sure a solution exists.
...

I don't think this is the case. In order to eliminate bots for a game-based blockchain you need to design a game that meets two criteria:

1) Bots can't "win" the game, or if they can earn some coins, it has to be many orders of magnitude less than a human playing the same game, otherwise someone can just run thousands of bots, right? You basically need a game version of the Turing test.

2) You need to have a game that, although bots suck at it, the results must be able to be easily and quickly verified by a computer. This is another wrinkle that makes point #1 much harder to achieve.

Afaik, a solution does not exist, and if you came up with one you'd probably be some kind of laureate or at least have a cushy CS professorship at MIT or something.

I can suggest these devs can combine a joint huge bounty to hunt for open source solution to feed these concerns. A lot of people will compete.

...

... The more you know, the more you can make strategic plans that would never occur to a bot. ...

Sorry I didn't answer this earlier... I was on a mobile device, and trying to do a lot of things at once.

I think I'm willing to hold the bounty, actually. The thing is, I feel that you underestimate the dev work required to make such a game bot proof. Even 1 BTC would be a relatively small compensation for the task. We will need to see vastly more support before anyone bothers to claim it... I'm curious how many other people echo your feelings, though