Thread by @dkcrypto13

Super busy today, but felt like I needed to put out a few thoughts on the #solana DDoS / downtime, how it was solved and what it means for $SOL & #DeFi

Firstly, I understand that what happened was that too many bots tried to claim/buy a token and spammed the network excessively.

There was already a patch that was released to take care of situations like these, but it was not adopted fast enough.

To solve this, the validators decided that they would do a re-start of the network with the new patch and it was fascinating to be a live spectator on discord.

Pausing here for a second, this means that, we have learned that #solana is indeed a decentralized network. It was impossible for "engineers" to go and solve the problem. They needed 80% of validators to take part in the re-start. It took almost a day because of decentralization.

That is a positive. The way the validators helped each other on discord and the transparency of the entire operation was also very welcome.

The fact a decentralized network got back up so quickly and is already back to smooth operations is commendable & it made #solana stronger.

There is, however, a major underlying issue and it will take a lot of brainpower to solve this. This affects not only #solana, but all of #DeFi. It was described very eloquently in this thread:

https://twitter.com/cometshock/status/1437870278684590091?s=21

In #DeFi there is a "liveness" assumption of the underlying network "baked in" to all lending, trading & staking. In other words, every protocol assumes that the chain it runs on AND the oracle's it queries are always available.

Furthermore, to make any borrowing possible, #DeFi uses overcollateralization as a mechanism. It basically means you pay in assets to a platform and you get to borrow 100% or so on top. Then you can take the 200% denominated in a different token and use it to generate yield.

The smart contracts you pay into assure that, if the total value of your 200% (the 100% you paid in and the 100% you borrowed) goes below 120% (or so), making the loan at risk, you get liquidated and the person that lent you money always gets back his 100%. You take the loss.

This is a brilliant way to enable borrowing and lending in a trustless setting. You do not need to check the creditworthiness of your counterparty as you know you cannot go below your principal if you lend money out.

The problem is: this only works if liquidation is possible.

In a scenario where the network goes offline and the oracles are offline too, nothing will happen while this is the case.

If you then have the network coming "live" again a day later and prices of almost all #cryptos have dropped dramatically (not exactly an unlikely scenario),

loans will all of a sudden be under-collateralized. Depending on how deep this goes, it could eat up the entire principal (and theoretically more!) and it would mean absolute calamity to everyone involved. Liquidations below par would happen everywhere, token that are staked as

collateral on other platforms would be worthless, in short: #DeFi would entirely collapse and you can be certain that an overcollateralization mechanism like that would never again gain the trust of investors.

We escaped this scenario of hell because prices were stable or up.

Had #Solana come back online with prices down 30% or more, we would have seen potentially the end of #DeFi. And this would mean its end on any chain.

It's not possible to convince investors that "your chain can't go down". Because all of them can. Or become too slow.

So what does that mean for #solana and #DeFi?

We dodged a MAJOR bullet.

We also cannot assume that this won't happen again. After all, bad actors now know that you CAN DDoS #Solana, so maybe they try a different route again in the future. What to do?

Suggestions:
1) Every validator needs to confirm/prove a 24/7 technical support or buy it from somewhere.
2) Every update to the validators needs to be compulsory within 12 hours.
3) We need to realize that this is a FATAL flaw in #Defi overcollateralization.

So projects and layer 1 teams should work on a back up plan. I am not technical enough to know what is possible, but basically I can think of:
- have a virtual "back up" liquidation engine that can liquidate using a centralized oracle when network is down

- use a cross chain solution as a back up oracle and liquidation engine (ie stay decentralized)
- Come up with protocols to follow in case of undercollateralization occurring (maybe assets should be locked for a while to see if they recover)
- Route via devnet (if possible)

I am not sure what can be done. I am obviously not a coder.

But I know one thing: this could have been #DeFi's death knell and we need to treat it as such as momentous event and try to come up with improvements that deal with such a situation. We all in this together frens.

Hoping the gigabrains at @mangomarkets @solana @iearnfinance @AaveAave @chainlink @PythNetwork @port_finance @autofarmnetwork @VenusProtocol @RaydiumProtocol @Solfarmio and @ethereum will all work out a way this horror scenario can be avoided. Thanks all :)