Comments on DSHR's Blog: Permacoin

A couple of links for those readers who still thin...

2014-08-22T10:23:47.713-07:00

A couple of links for those readers who still think the edge of the Internet is made up of desktop systems with lots of spare disk capacity.

A recent Bitcoin theft reveals another threat to n...

2014-08-07T14:54:58.675-07:00

A recent Bitcoin theft reveals another threat to networks such as Permacoin in which storage is rewarded with valuable "coins"; BGP hijacking.

Of course, even storage networks that don't reward with valuable "coins" are vulnerable to BGP hijacking, but there is much less motivation to employ it. Storage networks should validate both ends of each connection with SSL certificates using keystores supplied out-of-band, not via "trusted CAs" which are not in fact trustworthy. LOCKSS networks can be configured in this way.

The Bitcoin mining arms race continues apace. At l...

2014-08-01T17:12:08.821-07:00

The Bitcoin mining arms race continues apace. At least two mining silicon vendors have build 20MW data centers full of their mining chips in an illustration of the need for scale even to generate steady rather than increasing returns.

Industrial miners are expected to invest at least $100M/month over the second half of 2014. My estimate was too low by a factor of 6, so we are talking $300M/PB/yr.

This level of industrialization would make Permacoin essentially identical to S3 just 300 times as expensive. What exactly do you get for the extra $299M/yr?

Ghash.io is promising to be good in the future. Wh...

2014-07-18T06:20:15.113-07:00

Ghash.io is promising to be good in the future. What such promises are worth given their history is left as an exercise for the reader.

The unfortunate side-effects of storing "data...

2014-06-29T09:24:47.774-07:00

The unfortunate side-effects of storing "data of great public significance" in a system like Permacoin need to be taken into account too.

Henry, one misconception and one good point. What...

2014-06-27T15:03:19.307-07:00

Henry, one misconception and one good point.

What you want on your cell phone is your Bitcoin wallet. All that is in your Bitcoin wallet is the cryptographic credentials that secure your transactions; the coins themselves are entries in the blockchain which the miners store. Permacoin wallets would not be significantly bigger than Bitcoin wallets.

The storage in Permacoin is associated not with wallets but with mining. No-one mines Bitcoin on a phone except via malware (and Permacoin would make even that pretty much infeasible); these days it takes a rack full of custom chips to be a competitive miner.

On the other hand, you are correct that to be a competitive Permacoin miner, as MJSPK demonstrate, requires the miner to store its shards of the dataset in RAM (see Table 2) coupled with (at least) a GPU.

In effect, Permacoin is a storage system that keeps r copies, say 4, of the dataset in RAM. That is extraordinarily expensive in hardware and power compared to more conventional storage techniques. Couple that with the very rapid depreciation of mining hardware and it seems clear that Permacoin would be vastly too expensive to compete for the long-term storage market.

If my estimate of, say, two hundred million dollars per year for the rate of equipment investment in Bitcoin mining is correct, and we assume the RAM is free, and the network could store 4PB (see Section 7), we end up with $50M/PB/yr. Storing all 4PB in S3's Reduced Redundancy Storage costs just over $1M/yr. So even if Permacoin were successful it wouldn't be even close to cost-competitive.

David, A few other comments after some more tho...

2014-06-26T04:23:16.324-07:00

David,

A few other comments after some more thought.

1. The difference between Bitcoin and Permacoin is that amount of space needed is far different. You can move a bitcoin to your cell phone but only part of a permacoin will be able to be moved.
2. Permacoin depends quality storage vs. high performance CPU.