Very exciting to see the progress and the network mechanics being fleshed out! What you are describing makes sense.
Notes/comments:
If there was a standardized software package that frontends could install which performed checks on each video played by the frontend's users, then that could act as a checking mechanism. The software could have a standard way of checking the time it took to play the video and the quality of its encoding. Results of these checks (all results or just more significant ones) could be logged somewhere in the network itself such that they are visible to all. So, if based on the frontend reports it looks like there is poor encoding quality, or long time to play, coming from a particular encoding or storage node, then the validator nodes could check those particular nodes. So the validator nodes would be checking other nodes based on performance. Frontends have incentive to give good experience to their users so a collusion between them and encoding/storage nodes seems unlikely. What do you think?
At one point, the question came up of what would happen if people who have SPK disappear. A potential way to address this could be what we did on Hive - the governance votes of an account stop having effect if that account hasn't participated in governance for a year.
Regarding "The hard part" - preventing double spend of Broca for video uploads. Could this be addressed by the Broca being paid in advance to an escrow account? The upload job could start only after the Broca has been paid to the escrow. After the upload is finished, the Broca is sent to the storage node.
And a little feedback about the video - the volume seems very low, it was hard to hear. I imagine most people would not be able to hear the conversation if they don't have a headset and turn the volume to the max. Don't know if you have a way to increase the volume.