From the talk, it sounds like it can be implemented, the question is how much work will it take.
As you said, it is bias thinking but, outside of that, is there much value in it. I guess as someone mentioned, perhaps it might help gaming.
More of a talking/selling point than anything else.
Posted Using LeoFinance Beta
It'd be a fair amount of work I would think, You could reevaluate witnesses in the process, do you have 20+1 still or do you add more or less to run the blockchain faster? You could have 60 nodes do one block each 30 do 2 each, or keep the 20 and have more randomly selected backups which would help distribute witness rewards more.
Of course you could just keep everything the same but there are options to change some stuff while you are in the area of changing block times and scheduling.
Really the code changes would likely be block times and how the system schedules blocks to witnesses and reward pool calcs perhaps, other than that..
Optimisation would be the biggest thing needed, getting everything to run as fast as possible to support 1sec blocktimes, Sync speeds would also need to be super fast, otherwise nodes may get into situations where they can never catch up.
It'll do wonders for the snappiness of Dapps and stuff having new 1 sec blocks. Gaming could very well benefit from it the most. It is a 'Nice and cool to have' feature with some tangible benefits but a lot of thought is also needed as well.