a) you don’t trust the people behind the first one (or the country they’re based in, etc)
b) the underlying technology of the new coin is more appealing (tx/s, smart contracts, fees, etc)
That said, this is all Wild West and trusting any of these vendors to actually do what they say seems very risky.
If you’re 80% collateralized and 1% is fu money, there’s no defense against going to 78% collateralized if the code allows it.
"stablecoin loses peg" is an absolute proof that it wasn't a stablecoin any more than tranches of CDOs made high-quality bonds out of junk.
https://www.gemini.com/cryptopedia/gusd-stablecoin-gemini-do...
In the end, it's statistics (assuming that 10 out of 100 similarly sized debts default, the expected value for such a CDO is 90% of the face value of all its debts) and the assumption there will not be any major market issues such as a bank run, a pandemic or war that suddenly increase the risk of debts going bad - or in this case, the stablecoin running out of the liquidity it needs to maintain the peg.
CDOs and all forms of coins are essentially bets.
Stablecoins ought to be exceedingly boring and non-sexy. The key requirement ought to be "obvious, abundant and solid collateral", and nothing else should matter.