Quote from: HunterMinerCrafter on June 05, 2014, 02:12:20 PM

Quote from: WilliamLie2 on June 05, 2014, 09:18:31 AM

How blocks will be mined? You seed the map only with already mined blocks but how will you be able to mine new ones?

Yes, as I said there are quite a few fatal problems with my original plan. I have two rough directions that I'm thinking now... either allowing the last N blocks to be used for seed (so when you win a new block you might not necessarily actually be securing the current block, but adding security to any of the last N blocks. This would, however, mean that people would need to wait N times as many confirmations to be confident in their coins... I still think this is fraught with problems, however, and I probably won't continue exploring it much further.

If you are mining new block and secure some transactions with it then we say that these transactions belong to that block. If you say that they were part of previous block but wasn't secured by it then this is just a game of words, because you are that guy who secures them and you deside which transactions will belong to previous block, not the one who mined previous block, so it is still the same as if you added these transactions to your block. Each block includes hash of previous block, therefore currently all previous blocks are used for seed.

Quote from: HunterMinerCrafter on June 05, 2014, 02:12:20 PM

Another path that I am looking down is using something along the lines of the proof-of-activity proposal to allow user to collect up proof of stake data, and mine against that data to create blocks. Stake data would have to carry some "freshness" metric to avoid replay withholding, and manual human miners might only be able to mine coinbase TX in their blocks (unless they accept a map reset) but otherwise I'm thinking that it might work.

I'm not quite sure what you mean here. If they mine only coinbase then this is just a method to distribute coins, not to secure the network, right?

Quote from: HunterMinerCrafter on June 05, 2014, 02:12:20 PM

Quote

Currently, each block gives you infinite set of maps, not 2^32 but infinite.

Ok, I was assuming people were "playing fair" on their timestamp, as well. (My bots don't do any timestamp manipulation, they just take the "***Work" block that the client spits out.) Even with the timestamp the block count isn't quite infinite (there's an limit on timestamp drift allowed by the network ofc) but this does make the search space so large that, pragmatically, it may as well be infinite.

You don't need to change time, you can just change your address for coinbase transaction. With some other tricks you can have even more insanely larger set of possible maps.

Quote from: HunterMinerCrafter on June 05, 2014, 02:12:20 PM

At launch we had a very insecure network that was great for human miners of all skill levels.
Right now we have an exceptionally secure network that very very few humans are able to mine on.
The goal is to keep the network as computationally strong as it is now, but with a possibility for (ordinary, sane) human miners to continue to compete for coinbases. (EDIT to clarify: I mean secure the network without just punting and adding hashing based PoW like HUC does.)

Yeah, it is exceptionally secure. And you can perform 99% attack any time you wish, sounds very secure.

Quote from: Spiky on June 05, 2014, 12:12:31 PM

And what about adding some computationally intensive randomness to physics (like computing many many hashes each tick and adding it to velocity). This will make bot's speed comparable to normal human playing speed. And without careful planning bots will definitely lose.
Along with 4x sized map it will knock out bots for at least a couple of months.

Remember that physics is calculated not only while playing but also when other nodes check blocks for proper proof-of-play. Right now I can check about 60 blocks per second in one thread on my Core i7-3770. With 4 threads I will probably be able to check 240 blocks per second but many people have much slower processors. With your proposal physics simulation will become much slower and to really harm the bots we will need to make it at least 100 or 1000x slower. This will make block checking very slow, on some computers it will take several seconds to check one block, some probably would check new blocks slower then they are generated, inital synchronization with network will take days or even months, botnets will perform spam attacks sending a lot of invalid blocks and nodes will use 100% of their CPU time to check these blocks.