On the latest #Eth2 mass-slashing and Antifragility
A thread

My man @nntaleb notes how hyperoptimization comes at the cost of fragility.

Fragile systems are prone to black swans: they are so fine-tuned that when something goes wrong, they just blow up.

How do we get antifragile systems?

Usually resilient systems at a macro level are somewhat fragile at micro level.

This fragility comes from the fact that a multitude of approaches can be tried: some fail and get replaced by others that if they are right will remain.

The trial-and-error at the bottom makes the whole more stable

Decentralized networks apply this principle: a variety of client implementations, of node running setups, of geolocalization and political influence... having more of all of this makes the network more resilient:

If AWS goes down - some nodes go down but not the whole system.

Even if an entire continent got wiped out due to a bizarre electric blackout, nodes in other parts of the world would keep the network running.

Ethereum 2's design acknowledges that and wants to actively encourage it: by applying very small penalties to being offline, it encourages people to run it in non-professional settings.

At the same time, it is ruthless with potential threats

And yet here we are, chasing tiny increases in revenue/profitability, at the cost of jeopardising the whole system.

Doing things like eliminating the slashing protection because it has high I/O load

Why do I say "jeopardising the system" though? Let them be, right?

Yeah sure, but no thanks. Imagine everyone was as greedy for a as they are - i mean, obsessed with performance.

Imagine everyone was chasing this 1% more. Which I don't even know if it's true because the time scales are different, they could be cherry-picking time periods.

Everyone is hyperoptimizing, using the same ultra-sleek, no protections (they only slow us down!) setup.

And something happens. It all breaks.

Ethereum *actually goes down*. Price tanks. End of the party for all. Tears. Peepz jumping from tall buildings.

Fun fact: @ethereum researchers took this into account!

There is a mechanism designed to *discourage* the scenario above by applying anti-correlation penalties
Here, explained by the one and only @VitalikButerin : https://notes.ethereum.org/@vbuterin/rkhCgQteN?type=view#Slashing-and-anti-correlation-penalties

TL;DR: If you get slashed at the same time as others - you might be "correlated" - in the same pool, or using the same software with the same bug... you get triple fucked.

So: Don't worry, you are fine if your internet goes down or your power fries your machine.

You are still in time to help secure the network.

It's not complicated to do (shameless plug: @DAppNode ) https://medium.com/dappnode/step-by-step-staking-for-dappnode-users-92fdf7db0d0d

and repeat after me:

I
WILL
NOT
USE
FAILOVERS

unless you're talking to @luguslabs .