( 29th September 2019 )
Manufacturing Centralization
Bitmain does indeed produce a majority of double-sha256 (proof-of-work hashing algorithm behind BTC, BCH and a few others) mining equipment and the majority of hash power on the Bitcoin network as of this writing come from miners manufactured by Bitmain. We suppose here that though Bitmain may manufacture a majority of the mining equipment, they don’t necessarily control the equipment.
Fig-1: Manufacturing Centralization
Let’s discuss the risks of a single manufacturer producing most of the equipment used to secure the network through the below scenarios:
1) Backdoor
In this scenario, we assume that Bitmain sells most of the equipment that they manufacture. The risk here is that Bitmain has put in some sort of backdoor to the mining equipment through some hidden hardware, firmware or software. Some possible things a backdoor could do:
Force the mining equipment to point to a Bitmain-controlled pool mining whichever coin Bitmain chooses.
Override the block template to give rewards to a Bitmain-controlled address.
Throw away valid proof-of-work unless the miner is pointing to a Bitmain-controlled pool.
Shut down the miner using some predetermined signal. (Kill-switch)
The first two possible back doors would be really obvious to anyone paying the least bit of attention. Furthermore, they would be really obvious to prove with even a little bit of logging. The consequences of the discovery of a backdoor of this magnitude would pretty much destroy Bitmain’s reputation as a company and additionally make it the target of a class action lawsuit at a minimum.
This would be the equivalent of a kamikaze attack on Bitcoin, which might hurt Bitcoin short-term, but would completely destroy Bitmain as a Bitcoin ASIC manufacturer.
The third and fourth can be done in a subtler way, but would still be susceptible to discovery. Throwing away proof of work has the effect of delaying blocks while making the non-Bitmain pool look unluckier. Using a kill switch disables the equipment. The direct effects of both are actually detrimental to Bitmain as they have to deal with customer refunds and complaints about their miner not working. All benefits would have to be secondary, such as attracting more people to their pool. They also carry enormous risks as competitors might discover such backdoors and utilize them to Bitmain’s detriment. The risk of discovery alone more or less destroys whatever reputation Bitmain enjoys and the benefits uncertain and far off.
Note that these are things any mining manufacturer could put into their equipment. Hardware is very hard to audit and by buying equipment from a particular manufacturer, you’re in a sense trusting them not to cheat you.
In the 4–5 years of existence, Bitmain hasn’t resorted to these tactics, and there’s no reason to think that they would. Such backdoors involve a lot of planning, a high chance of discovery and/or failure and a low chance of reward.
2) Manufacturer Defect
This scenario assumes that everything else is the same as scenario 1, but the equipment has some fatal defect. Perhaps the equipment catches on fire above a certain temperature. Perhaps the equipment calculates the timestamp wrong.
The worst possibility here is that the equipment creates invalid blocks and that’s easily seen by the rest of the network. Again, this only hurts the manufacturer, as they are the ones that have to deal with their customers’ anger.
3) Price Gouging/Buying Restrictions/Shipping Delays
In this scenario, the equipment manufacturer uses its dominant position to add additional costs for buyers of the equipment. The costs may be charging more for the equipment, forcing the usage of certain payment methods, delaying shipping, perhaps even restrictions on how the equipment may be used.
All of these tactics become intolerable under competition as the total costs of the equipment can’t go above that of the competition without hurting sales and thus must be used judiciously, if at all. The additional revenue from acting this way is offset by long term reputational damage.
Comments