These are chat archives for jdubray/sam
The key sentence is:
"Personally, I think all distributed computing depends on timeouts, retries, and idempotence."
Idempotence being the elephant in the distributed room. That's my I believe SAM has a much better chance than a pure function approach because the tuple action/model/state offers a much better way to create idempotent operations than a big fat reducer flanked by anemic actions and stores. Immutability doesn't help there a bit.