All software has bugs. This is same as claiming ETH will always get hacked because of the DAO hack.
As far as I can tell there is no fundamental design flaws in Solana that makes it so that it will always go down. There are bugs in the networking stack that are being fixed.
An no, "engineers" don't bring it up. I run a validator and only i'm incentivized to restart it back up because of my stake.
This is no different from ETH, Lido staking has ~30% market share[0].
Stake centralizing != less security. no amount of stake can allow validators to steal your funds, this affects only censorship resistance. So actual validator count is more important than superminority.
In case of USDC I'm assuming Circle should be running their own validator, which is the source of truth.
As someone who has used most networks, Solana has the best experience for transacting USDC.
I've tried doing this read & updates using Postgrest and row-level-security using Supabase. When it works its an amazing experience but even for semi complex stuff you would still need to use Postgres RPC, which is still another "API" layer. I find writing API using SQL a nigtmare.
Simple queries like this don't work on Postgrest:
`update likes set likes = likes + 1;`
- use coinjoin with something like wasabi wallet(https://wasabiwallet.io/)
- purchase BTC with cash