Mining is Centralized! What now? (Trust and Blockchain Marketplaces)

by David Vorick
Mining is Centralized! What now? (Trust and Blockchain Marketplaces)
·
connecting...
Contributors (2)
D
Created
Oct 05, 2019

Real-time transcript.

Summary

Miner centralization is inherent

There are only 5 main orgs manufacturing rigs. Mining mfg is more of a winner-takes-all. There aren’t mutliple features: you want to do a lot of hashing at low electricity.

Bitmain, now Whatsminer

Mining pools

8 mining pools have 80%? of the hashrate. protocols let them choose what blocks get mined; there’s work to try to fix this. The pools are more flexible; if you decide you don’t like bitmain you can’t pus them out; but if you dislike a pool you can spin up a new one / write new software.

Mining farms

It’s hard to know what mining farms exist and wher they’re located.
It’s hard to know how you would even try to decentralize e.g. bitmain and their datacenters. Efforts to decentralize have failed, often at great expense.

Inherency of centrality

Mining is zero sum. A new competitor makes everyoe else make less money. New products are value-shifting, not value-add; always drawing from existing frameworks. It’s a huge ongoing expense, so anyone can push you out.

Worse, any funds spent trying to ‘decentralize’ is not being spent on optimizing for zero-sum survival. Unethical mining pushes out ethical when it’s more effective.

What if we assume monopoly miners?

Assume a given system can’t avoid having collusion among a majority of miners. You essentially have a single monopoly miner. What then?
Claim: I think with research we may find monopoly miners are okay for most blockchain applications.

1) Limit miner powers, and window of opportunities for malice.

2) Add a penalty for causing damage / going rogue

Miner powers

Fundamental powers:

1) Miners can censor transactions
2) Miners can reorganize history at a cost

Bonus miner powers:

3) Can confuse SPV nodes
4) Can set block fees + other parameters?
5) Can unilaterally change consensus rules

Avoid these. Don’t let minors change consensus; don’t endorse ‘SPV’, keep parameters? separate

Limiting censorship

Blinding: make it harder to distinguish transactions from different actors. Make transaction privacy a chain security parameter.

Limiting double-spend potential

Use higher confirmation rate for large transactions.

Hand-wave: Use layer-2 channels buried under months of work for as many transactions as possible.

Penalties

Detect attacks:

Censorship is seen when txns with appriraiate fees aren’t mined.
History re-writing is detectable when a reorg reaches a statistically unlikely size.

Penalize miners:

Penalties have to be off-chain.
Identify and have recourse over miners (?) high startup cost to be a miner…

  • Walk away altogether. Leave a blockchain ecosystem; sell all coins and devalue? [hm]

  • Change hashing algorithm. Have a supply chain ready that mines the new algorithm. [cf Sia’s change. effective? cost miners $100M in total?]
    But: inefective against supply chain attacks.

  • Counter attacks: if a successful attack would lead to massive chain-flight, the market ap of the chain may be incentivized to counterattack? (against commodity hardware) [hm]

Questions

Can you tell in the system what time a txn is posted? If so why not add a penalty for leaving txns off over time?
Yes, but causes complex interactions w/ reorgs. If txn fee drops over time, you have incentive to pretend you got the txn earlier in time.

Comments
3
Samuel Klein: Does this ever happen? When? Don’t ignore cost to leavers
Samuel Klein: Are client-designers/coders separate from miners in this thought experiment?
Samuel Klein: Who defines this?