YashasEdu(@YashasEdu):I personally align with this take. Because I think he race for frontier intelligence will be about whether the system that produces and distributes that intelligence stays open + permissionless or whether it ends up controlled by a few closed platforms that can ration, monitor and shape it and not about which model leads the benchmarks for the next 2-3 quarters Blockchain is still the only tech we have that can coordinate capital, compute and contribution at scale without needing permission from any gatekeeper and opensource AI at the frontier needs exactly that kind of system. TCP/IP buried a dozen rival protocols we no longer remember, yet the internet still arrived. The same pattern is happening again but this time the stakes are not about money. A world where the intelligence layer is closed by default becomes a surveillance system by design and the West does not win by building higher walls. It wins by shipping the open alternative and letting the rest of the world use it. That alternative is already live... 1. Subnets keep competing to reduce loss, compress models and deliver useful intelligence 2. Hundred-billion parameter work is being trained in the open 3. Performance auctions settle in real time 4. Hospital GPUs are already earning while the machines sit idle overnight Intelligence is starting to look like a resource you can simply draw from a socket, no matter which lab or country produced the latest improvement. Also the incentive design is still being improved, just like every early infra layer but the core system is running, compounding and already delivering real output that closed systems cannot match in openness. For me @bittensor bittensor:native is the clearest working version of that open infra right now.

2026.08.10 04:38

I personally align with this take. Because I think he race for frontier intelligence will be about whether the system that produces and distributes that intelligence stays open + permissionless or whether it ends up controlled by a few closed platforms that can ration, monitor and shape it and not about which model leads the benchmarks for the next 2-3 quarters Blockchain is still the only tech we have that can coordinate capital, compute and contribution at scale without needing permission from any gatekeeper and opensource AI at the frontier needs exactly that kind of system. TCP/IP buried a dozen rival protocols we no longer remember, yet the internet still arrived. The same pattern is happening again but this time the stakes are not about money. A world where the intelligence layer is closed by default becomes a surveillance system by design and the West does not win by building higher walls. It wins by shipping the open alternative and letting the rest of the world use it. That alternative is already live... 1. Subnets keep competing to reduce loss, compress models and deliver useful intelligence 2. Hundred-billion parameter work is being trained in the open 3. Performance auctions settle in real time 4. Hospital GPUs are already earning while the machines sit idle overnight Intelligence is starting to look like a resource you can simply draw from a socket, no matter which lab or country produced the latest improvement. Also the incentive design is still being improved, just like every early infra layer but the core system is running, compounding and already delivering real output that closed systems cannot match in openness. For me @bittensor bittensor:native is the clearest working version of that open infra right now.

Aylo@alpha_pls

2026.08.09 17:51

I've never returned to a project as much as Bittensor. I see the promise, I see the rise of open source ai, but every time I think there is momentum I am reminded that there is a centralised committee endlessly tweaking monetary policy. Getting this right means calibrating TAO's monetary policy, the system mechanics, and subnet incentives simultaneously, so grifters are priced out, real builders are paid and TAO holders are not subsidising extraction. This is a very hard problem. There has been a lot of emission wastage so far, but there is a persistence to get it right, and it is remarkable they have even come this far. The dedicated cult is strong, and bears a similarity to early Ethereum. The subnets are a bit like the early DeFi projects that looked rough around the edges/totally implausible (a few look better than that). Like a moth to a flame I do return to see how the subnets are doing, and what they have produced, and there are continually results of note, and that is unique in crypto. It's intellectually stimulating, unlike the other dark side of our speculative crypto spectrum (memecoins). One subnet breaking out reprices TAO, which is another reason why I can't just dismiss Bittensor, as some do. I would be more surprised if this didn't happen at this stage. I have gone through many periods convincing myself that Bittensor will never work, but then I have a look again a few months later and my monkey brain sees the bull case (especially for the next cycle). There are still far more reasons why Bittensor will fail in the long run than succeed (this is always the case with early stage innovation), but at least there is a very big ambitious vision that looks like it would add a lot of value to the world.

커뮤니티로 전달