Congratulations on all the progress in the past year and thanks Eli for the transparency and answering everybody’s questions here and on Discord.
My sense is that Streamplace is a large and ambitious project and you clearly have a vision for where things should go. That being said, I do have a few additional concerns:
The proposal, even the latest split one, could have used a little more time to marinate before being put up for a vote. Folks are still thinking through things and have raised many good points. Eli, for your part, you are doing an excellent job of being responsive to these concerns, but I am not sure what the urgency was in putting the proposal up for vote. That gives me pause.
Transcoding plays a small (but important!) role in Streamplace. As you’ve clearly explained, there is a lot that goes into a successful startup: the partnerships, marketing and outreach, development of core and adjacent technologies, etc. The Livepeer network benefit is second-order, and still speculative. Is the Livepeer community alone in shouldering the financial risk for these non-ecosystem initiatives? Streamplace seems to be as much (if not more) of an AT Protocol play - are there resources there for you to tap? Committed funding from entities in that sphere would be great validation - it doesn’t have to be of the same magnitude, but it’s a big vote of confidence.
Given the size of the ask, and that (all?) your funding is from Livepeer, I would like to see more ways to make use of the Livepeer network outside of transcoding. AI moderation would be a good candidate, even if that means development of new pipelines in order to be useful. That would be an excellent way to contribute capabilities back to the ecosystem and bring additional usage onto the network. Are there other areas where Livepeer might be able to plug in with its non-video capabilities, eg LLM or TTS / SST? What about distribution?
I get it, I wasn’t aware we were asking folks to let things sit for at least seven days. Mea culpa. That said, I see the next eight weeks as incredibly important; I’ve known the Skylight team for a little while now but nobody predicted it would pop off quite as hard as it did. I feel like I need to make something happen right now or the chance won’t come again. Thus the drop to a less-controversial amount and the immediate posting.
That’s a really excellent point - you know, it’s funny, my intuition had that entirely backwards. In my head I was like “I’m going to go back to the treasury first, of course, it’ll be respectful of me to offer right of first refusal.” So I didn’t make a big effort around that to this point, but you’re totally right, it’d do a lot to demonstrate the momentum around the project. Let me see if I can’t scare something up before the end of the voting period. Coming from a Web3 background you’d be appalled at the minuscule resources available for public goods funding in the AT Protocol universe, but I bet I can arrange for something.
That having been said — yeah, it’s an AT Protocol play. Bluesky bootstrapped the world’s most successful decentralized social network and the first wave of video applications is just now launching. You’re gonna want some ATProto plays.
Distribution is the huge one — if go-livepeer renamed its ingest functions to “gateway” so I’m happy to seize the unused word “broadcaster” for Streamplace. It’s been designed from the ground up for simple peer-to-peer replication, moderation, and embedded provenance. If orchestrators want to get into the content distribution game, this is the foundation, absolutely.
Text-to-speech and speech-to-text are a really great fit too - if the models are ready on the server and can be sent back with the MPEG-TS we could get something basic working for this in a day or two. We were talking about it today on stream with @rickstaa actually!
The AI moderation side is a big opportunity if there’s someone that’s able to do the work of getting the right models. That’s the part that seems really hard to me. If we can figure out that part, we’ll absolutely wire it up to the Streamplace moderation infrastructure absolutely. I’m trying to think of the kinds of models that would be a good first line of defense - adult content would probably be reasonably well-defined.
Eli said it himself, he got $360k in the first proposal. There was nothing to show in the beginning and he got 360k. Now there is substance to show, we see a lot of progress has been made and Inc is showing resistance. Why would one say OK to 360k then, when there was nothing to show but resist to the same amount now, when there is good proof of work in front of us?
Look, I know this looks like VC capital again. But I think it’s worth it. Most grants and treasury funds so far on Livepeer AI, as far as I understood them, was put to good use but all demand through them, all demand due to them, they are all incentivized. Livepeer still doesn’t have any real Ai revenue after 1.5 years. Pmf is still being explored. It’s still in trial/error phase, still in experimentation phase, still “let’s give money to folks to use Livepeer GPUs” phase. But with this, at least there are clear alternative paths forward in terms of pmf. At least it is already known how real money can be made thru this. It might not work out but it’s better to give it that chance than not giving it imo because Livepeer needs demand more than anything and this is the best candidate so far as far as I can see.
“Public goods”. Real demand is the best public good for Livepeer right now.. I don’t wanna see just code and no action. Streamplace does the code and seems to have found a real user/customer for it. To me, this is what “public good” is.