I had always been quite confident that the “each instance maintainer will do moderation” model would be an abject failure (moderation at scale is hard, but still easier than individual moderation), but I’d assumed that mainstream instances would react by not moderating, leaving space for better mod models. Instead we have many instances that black-hole HUGE swaths of “the fediverse”. Including the amazingly anti-federation practice of blocking mastodon.social.

Not fit for purpose.

The obviously “correct” thing is to set interest at 1-(1+base)(1+credit): base of 5%/10%/20% at credit of 100% (or 5.26%) gives rates of 110%/120%/140% (or 10.5%/15.8%/26.3%) which all return *exactly* the base rate on average.

It’s just very odd to me that even sophisticated borrowers can’t be sold on this not-very-complex formula instead of the naive add-the-rates approach.

But okay; assume banks “price in” that premium, so high credit rates are nudged even higher. The whole point of the “base+credit” system is that credit is fixed while base (and total) float.

But in our first example, if base goes up to 20% the return is $110 (a $10 loss to base) and down to 5% it’s $102.5, a $2.50 loss.

Merely adding the two rates means that the bank really cares what happens to the base rate. And I thought the point was that they wanted to be indifferent to it?

Marginally more realistic numbers: 5% base; 1/20 chance of default meaning a 5.26% credit rate. Nineteen times out of 20 a $100 loan returns $110.26; 1/20 nothing; return $104.747.

The fact that this is below the base rate isn’t what bothers me nearly as much as that /how much below/ is a function of the credit risk…and that *higher* risks are *less* lucrative for banks.