Some articles that I have bookmarked:

• https://medium.com/@factchecker9000/nothing-is-worse-than-proof-of-stake-e70b12b988ca
• https://tuurdemeester.medium.com/critique-of-buterins-a-proof-of-stake-design-philosophy-49fc9ebb36c6
• https://medium.com/hackernoon/sharding-centralizes-ethereum-by-selling-you-scaling-in-disguised-as-scaling-out-266c136fc55d

from what i recall:

• you could lend money to control the chain.
• you need to have the stake online to prove.
• single point of failure while in pow you need silicon and power.
• calculation wo can mine the next block has issues.

The one model that I think is useful for evaluating the trust required is
https://vitalik.ca/images/trust/trustout.png

Where would proof-of-stake fall in there? Is there difference between different implementations?