This really isnt the place for this.. but...
This is good place for this, since I'm sure this solution also will solve spam problem. But its primary point is to build scalable, distributed, easy-accessible block chain. I'm sure if it will be built, its capacity will allow to don't take attention to any spam.
Pruning does not break chain integrity. From my somewhat limited knowledge about it, pruning just removes the body of a transaction from a block and leaves the header and its hash alone. As a result blocks can still be verified. Each client would do the pruning themselves, at which point if no copies of a old transaction are floating around then its safe to assume that everyones client removed it, it was no longer needed. Regardless im more then sure there will be a few complete unaltered blockchains around if anyone needs to look up such data. Perhaps thats something your model would work for.
I absolutely don't accept any items in chain to be abandoned. Its a history of evolution. All items in chain must be stored
and accessed easily from any point of Internet. This is the main point of DHT storage: scale chain and make it easy accessible. As secondary benefit I see disappear of spam problem (disappear not spam, but problem) and ability to run full chained client on any cheap or expensive hardware.
Regarding IO capacity of HDDs, i cant say because i dont know. But it doesnt seem to be a problem yet. I also think there are more pressing issues.
This. I do know what is IO problems. They hide for a long time and appear instantly choking project in bottlenecks. My primary work is to build distributed storages for heavy random IO operations. I know what is when SATA HDD allows only 10 MBit/sec throughput on random reads. I ask you, all don't put the whole network load on one regular client!
Maybe you should write up a paper about it and see if you can get the devs to look at it and tell you if your going in the right direction.
Probably yes. I will develop one.
