What Root Network weights reveal about Bittensor

In this article, we look at how the biggest validators in terms of TAO delegated set their weights between subnets.

Currently, validators on the Root Network (top 64 validators) set weights on subnets to determine how much TAO emissions they receive (set to change with BIT001).

Directing emissions is key in Bittensor as it is what shapes the growth of the network. Depending on where they put the emissions, validators could be pushing promising subnets forward, or throwing TAO out the window.

Everyday, these validators have the power to distribute 7200τ, more than $2m dollars at current prices (~$300 per token).

Now that we all realize how much it matters, let’s have a look at what’s going on behind the scenes!

How do top validators set their weights?

Before getting to it, it’s important to realize how concentrated this process currently is: the top 5 validators direct more than 65% of all TAO delegated. If we add the 5 next validators, we climb up to around 85% of all TAO delegated to only 10 validators.

In other words, what this group does is what moves emissions in Bittensor. Here a the charts:

Opentensor (25%), Taostats (13.5%), and Foundry (11%)

Subnet 0 is at the bottom

Bittensor Guru (9%) and RoundTable21 (6.5%)

The first striking thing is that SN0, recycling TAO*, is the most weighted subnet by 8 out of the top 10 validators.

*weights directed towards subnet 0 are recycled: they don’t get issued now and will be in the future.

Except for Tao Validator and RoundTable21, all the other validators have put more than 10% of their weights towards recycling.

Datura (6th), Bittensor Guru (4th) and 1stTensor (8th) respectively put 100% (!!), 70%, and 65% of their weights on subnet 0!

Which brings the question:

Why are validators putting so much weight on recycling TAO?

We see two potential answers for why validators prefer “saving” the emissions:

1. There simply aren’t enough valuable subnets for all the TAO issued.
2. Validators are having a hard time looking at all the subnets and evaluating them.

1. Is there a lack of valuable subnets?

When TAO went from $50 to $700, subnets revenues 10x in less than 4 months. Some subnets adapted and improved drastically, others couldn’t keep up and died. Most survived, but were not worth the incentives they were receiving anymore.

This gap between funding and valuation could explain why some validators prefer “recycling” TAO emissions for the future, when more promising subnets might emerge.

Another explanation is that validators wanted to try to slow down the decrease in TAO’s price.

Looking at subnets from another angle, they are net sellers of TAO. The tokens they receive are funding for miners, validators, and subnet owners.

In that sense, the tokens that they receive should be expected to be sold. Some see that as it is: bad for their bags. Which explains why they started to push this recycling narrative.

But the selling of those tokens is what motivates value creation in Bittensor. Without incentives, there can’t be any value creation.

It’s a chicken and egg problem: no one wants to start producing value unless there are incentives to do so, and no one wants to incentivize projects that don’t produce value … unless?

Unless they have skin in the game. And this is coming with Dynamic TAO.

Before getting into that, let’s explore our second theory: validators are having a hard time evaluating subnets.

2. Is it hard to analyze every subnet?

To answer that question, we can look at how many subnets validators actually put weights on on average. Out of 39 subnets at the time of writing, what would be your guess? 35? 30? Twenty-one.

On average, the top 10 validators (85%+ of all TAO delegated) distribute incentives to only 21 subnets.

Removing subnet 0, 3 validators out of the top 10 weight less than 10 subnets.

Additionally, 2 out of 10 validators (15% of TAO delegated) weight all subnets they choose equally: they just select which subnets they think are good enough to receive TAO emissions, but don’t evaluate how much and put the same weight to all of them.

Bittensor Guru and 1st Tensor

This shouldn’t come as a surprise. Understanding how each subnet works, how serious the team is, and if it’s worth the incentives is HARD. And it’s only going to get harder with the number of subnets increasing.

Unfortunately, Bittensor validators need to do a thorough analysis of every subnet if we want TAO emissions to be distributed efficiently, but it seems impossible … unless?

Unless many participants are incentivized to analyze subnets.

There are two key elements to be improved here:

• many participants: the gates of the TAO distribution need to be more open and accessible.
• being incentivized: participants need to have an incentive and a risk (skin in the game) to direct TAO emissions to one subnet or another.

Conclusion

Before we finish, let’s gather what we learned from the data:

1. The current mechanism is suboptimal at allocating all TAO emissions, as only 70% of the tokens are currently being distributed.
2. It’s very hard, if not impossible, for a single validator to evaluate all subnets.

Are you thinking about it as well?

This is exactly what Dynamic TAO brings to the table.

Dynamic TAO will let every TAO holder decide to distribute incentives to one subnet or another, and tie their returns to the success of these subnets.

One aspect important to remember with DTAO is that it won’t get easier to get an educated opinion about all subnets, but more probably harder. Instead of trying to analyze all subnets, TAO holders and delegators’ goal should rather be to focus on a few subnets, and direct incentives where they see the most value and potential.

This is what we will be doing at Mentat Minds, for us and for any TAO holder that want to.

If you’re interested, you can get in touch on X or at team@mentatminds.com.