ການຄອບຄອງ Prysm ຂອງລູກຄ້າ Ethereum Staking ມີຄວາມສ່ຽງຕໍ່ການລວມຕົວ - Trustnodes

Run by grant funded Prysmatic Labs and written in Go, Prysm began dominating since the very beginning on testnet.

Hopes that would change once mainnet launched have not materialized, with now more than a year later Prysm still accounting for 62% of all eth 2 clients.

That’s according to data by Miga Labs run by Dr. Leonardo A. Bautista Gomez (Leo Bago), Team Leader and Senior Researcher at the Barcelona Supercomputing Center.

Until now this data was hard to come by, but their crawler presents some nice charts, including the up to date featured image above.

That shows with just a few changes, the network can become resilient. If just 10% of Prysm clients go to Lighthouse, for example, and 20% to Teku, then all three would be below the 33% threshold.

A threshold that is lower than in Proof of Work where 51% is required to be malicious, with the definition of ‘malicious’ in Proof of Stake (PoS) including simple accidents like some innocent bugs in Prysm.

If there was such bug and it was exploited currently or causes the client to go out of sync, then the whole ethereum 2.0 network would stop running for potentially days or even weeks as stakers move to other clients or wait to be ejected, losing tons of money in the process.

No such thing has happened on the live network, but it did on testnet in 2020 when it crashed due to some client bugs with it taking quite some time to get back running.

One way to guard against that is to diversify clients so that no one client has more than 33% share of the network, a bit like no one mining pool should have 51% of the network.

In which case if there is some problem at one client, then its users may face difficulties, but the network as a whole remains unaffected. Otherwise if that client has 34% share, then the whole network basically stops.

To prevent that, there have been attempts to make transferring of staking between clients easy, just as there have been attempts to persuade big staking providers to diversify.

A Coinbase employee in their staking department, when asked, said they’re working on diversifying their own staking nodes, with the data above showing there has been a slight drop in Prysm from 65% share of the network in December.

Yet the dominance is still so high that it might affect plans to launch the Merger, that is the full network upgrade to Proof of Stake whereby everything becomes subject to this 33% threshold.

Currently the running Beacon chain network doesn’t have transfers, so any problem would be limited. Once the Merger goes live however, all the collaterals and bots in Aave or Dai and pretty much everything running on eth would be subject to the new Proof of Stake mechanism.

So a small bug in Prysm, if it is still as dominant by that point, may bring everything to a halt which could be a bit like Bitmex going down amid volatility as once the network is back up and running, the oracle feeds would have different prices you’d think.

Which is why some devs are not comfortable launching the Merger while Prysm still dominates.

The launch itself may be this summer if all goes well in the Merger testnet, and if the 33% client threshold becomes a requirement for it, you’d think it would easily be achieved as stakers would want it to launch.

However, once it is launched then we may still get this dominance afterwards because stakers are clearly liking very much this Prysm client.

So there might not be a forceful solution except for relying on staker’s good sense and self interest to not lose money, in addition to pestering companies like Coinbase to diversify.

Seamless transfers of staking clients would also be useful, and may even solve the problem completely as then you just have a backup client that would make diversity pretty much inbuilt.