DPoS consensus used by Bitshares/STEEM & EOS seems to the most advanced among the models and they also have the highest TPS
- Finality Latency: how fast is a block finalised after it’s been proposed
- Decentralisation: how many consensus-forming nodes can exist in the network
- Resource Overhead: how many blocks per second do consensus-forming nodes need to verify
dPoS protocols sacrifice decentralisation and compromise in resource overhead for apparent performance, ie many transactions processed by a few very powerful servers.
The game-theory and security aspects of a blockchain like its censorship resistance, immutability or who can produce the next block are also relaxed to a level that many blockchain professionals consider unacceptable. Case in point the formation of cartels is incentivised by the dPoS delegation of vote mechanism. Corruption is baked in.
I wrote a short intro to consensus protocols a while back if you're interested in knowing a bit more: https://medium.com/@jpa_of_snc/consensus-casper-and-cryptoec...
In any case, Bitshares TPS was less due to DPoS and owed more to the limitations of what tx were supported.
Ethereum chose to support greater contract flexibility which turned out to be what the community of users wanted more, at least then.
We've yet to see a real solution to the problem of having to tradeoff performance for decentralization (while keeping security.)