Soft fork fix for block withholding attacks

Here's a method of fixing block withholding attacks with a soft fork:

We require blocks to choose an arbitrary target, e.g. two bytes. We redefine the block PoW target to be "less than the difficulty, with the last two bytes being the target".

We require blocks to include a blinded hash of the target plus some junk (which blinds it) in the coinbase. The miner cannot arbitrarily switch targets.

The miner can now send the block header to hashers. Hashers do not know the target, and hence must submit all shares that matches the first PoW criteria (below difficulty) and delegate secondary verification to the miner. With two bytes as the target, there are 65335 false positives for every valid block.

Finally, we require miners to communicate a proof of their target hash (ie, the junk they generated) in a non-hashed area of the block. This can be a protocol extension. The target is already included in the hash as the last two bytes.

This can be deployed as a soft fork with miner opt in, triggering across many difficulty cycles. Initially, we soft fork to require the last bit of the block hash to be zero. The next difficulty cycle, we require the last two bits to be zero. We do this 16 times to get 2 bytes, and then we actually activate targets.

By then, nominal difficulty would have been cut by 2^16 (65536), but the block target makes mining each block 65536 times as hard. We do the progression over 16 difficulty cycles to minimise increases in block timings. We can be more specific and progress over even more difficulty cycles through more clever soft fork rules.

For example, Vitalik detailed "timewalking" attacks that allow effective granular lowering of the nominal difficulty.

Critique welcome.

On Fri, Feb 12, 2016 at 10:31:56PM +1100, gladoscc via bitcoin-dev wrote:

> Here's a method of fixing block withholding attacks with a soft fork:

So, while you're technique I believe works, it's not a soft-fork, at least under the definition most of the Bitcoin dev/research community have been using.

The reason is if it's adopted by a majority of hashing power, less than a majority of hashing power can create a chain that appears to be the most-work chain, from the perspective of non-adopting nodes. Those nodes would then be following a weaker chain.

A better term for what you're proposing might be a "pseudo-soft-fork", given that you don't quite meet the requirements for a true soft-fork. Having said that, it may be the case that overall your technique still reduces risk compared to a simpler hard-fork implementation of the idea; more analysis is needed there.

If clients were designed to warn their users when a soft fork happens, then it could be done reasonably safely. The reference client does this (or is it just for high POW softforks?), but many SPV clients don't.

If there was a delay between version number changing and the rule activation, at least nodes would get a warning recommending that they update.

* At each difficulty interval, if 950 of the last 1000 blocks have the new version number, reject the old version blocks from then on.

* Start new target at 255, the least significant byte must be less than or equal to the target

* Update target at each difficulty re-targetting

T = ((T << 3) - T) >> 3

This increases the difficulty by around 12.5% per fortnight. After 64 weeks, the target would reach 0 and stay there meaning that the difficulty would be 256 times higher than what is given in the header.

An attacker with 2% of the network power could create 5 blocks for every block produced by the rest of the network.