Bittensor opens up more competition slots for AI agents—but who decides their rewards?
OpenTensor recently introduced Null Consensus on X, linking to Runtime 475, released on October 7. It offers an interesting option: subnets can expand the size of competitions while concentrating miner scoring authority in a single evaluator. For developers, the opportunity and the constraints need to be considered together.
According to the current official guide, under a single incentive mechanism, Null has a capacity limit of 2,500 UIDs, compared with 256 for Yuma; when the two mechanisms are shared, each gets 1,250. A UID is a registered participant slot, not necessarily a unique user or paying customer. Subnets still use Yuma by default, and choosing Null does not automatically increase the participant limit.
What really shapes the competition experience is the reward rule. Under Null, the registered UID with the highest stake weight gets the sole validator permit and uses its effective weight to determine miner rewards. Other participants with a positive stake can still receive validator dividends proportionally. This gives more teams a chance to enter the same competition pool, but once they secure a spot, they still need to understand how this evaluator judges task quality.
My tentative view is that this is better suited to experiments that turn clearly defined tasks into large-scale competitions. A single evaluator could make standards more consistent, but it could also concentrate bias. Ultimately, task quality, evaluation transparency, and genuine paying demand all need to be assessed; capacity alone cannot answer these questions.
As of around 08:08 Beijing time, Binance TAO/USDT was up about 3.03% over the rolling 24 hours, with roughly $27.25 million in trading volume. The spot hourly candle that closed between 07:00 and 08:00 was down about 0.04%, with roughly $583,800 in volume—about 0.83 times the average of the previous five hours. The daily gain is not the same as the move in the most recent hour, and there is no evidence linking the price action to this announcement.
Deployment caveats also remain: the official release details still show “Upcoming,” and the indexed capacity summary does not match the guide. I have not independently verified whether it is enabled on mainnet. Next, I’ll check whether specific subnets have switched over, how many participants are active, and what revenue tasks generate. If the new slots remain empty over time, evaluations prove unreliable, or the rules are revised, the view that expansion will bring product growth will need to be reassessed. $TAO
OpenTensor recently introduced Null Consensus on X, linking to Runtime 475, released on October 7. It offers an interesting option: subnets can expand the size of competitions while concentrating miner scoring authority in a single evaluator. For developers, the opportunity and the constraints need to be considered together.
According to the current official guide, under a single incentive mechanism, Null has a capacity limit of 2,500 UIDs, compared with 256 for Yuma; when the two mechanisms are shared, each gets 1,250. A UID is a registered participant slot, not necessarily a unique user or paying customer. Subnets still use Yuma by default, and choosing Null does not automatically increase the participant limit.
What really shapes the competition experience is the reward rule. Under Null, the registered UID with the highest stake weight gets the sole validator permit and uses its effective weight to determine miner rewards. Other participants with a positive stake can still receive validator dividends proportionally. This gives more teams a chance to enter the same competition pool, but once they secure a spot, they still need to understand how this evaluator judges task quality.
My tentative view is that this is better suited to experiments that turn clearly defined tasks into large-scale competitions. A single evaluator could make standards more consistent, but it could also concentrate bias. Ultimately, task quality, evaluation transparency, and genuine paying demand all need to be assessed; capacity alone cannot answer these questions.
As of around 08:08 Beijing time, Binance TAO/USDT was up about 3.03% over the rolling 24 hours, with roughly $27.25 million in trading volume. The spot hourly candle that closed between 07:00 and 08:00 was down about 0.04%, with roughly $583,800 in volume—about 0.83 times the average of the previous five hours. The daily gain is not the same as the move in the most recent hour, and there is no evidence linking the price action to this announcement.
Deployment caveats also remain: the official release details still show “Upcoming,” and the indexed capacity summary does not match the guide. I have not independently verified whether it is enabled on mainnet. Next, I’ll check whether specific subnets have switched over, how many participants are active, and what revenue tasks generate. If the new slots remain empty over time, evaluations prove unreliable, or the rules are revised, the view that expansion will bring product growth will need to be reassessed. $TAO