The Constant is Change
Bittensor is not the same network it was three months ago, let alone a year ago. Kusanagi, Nakamoto, Finney, Revolution, Dynamic, Flow. Six versions of the same network, each one its own era defined by a rebirth, an expansion, or an optimization. Every one of them moved the network somewhere it had never been.
People complain that keeping up with Bittensor requires constant attention. But that's exactly how it should be. We are not optimizing for how easy the network is to follow, but for living at the bleeding edge of incentive computing and winning on that frontier. That means rapid experimentation, both successes and failures.
This past week, from where I sit, we crossed into a new era. I’ll call it Velocity. This one is defined by moving faster and making the hard, unpopular calls to get this network back on course. And successfully doing this requires centralizing the core direction of the protocol.
As an operator and investor deep in the ecosystem, I see the problems in the network, and the fixes they demand are heavy ones. Emission allocation needs to be redesigned, validators need to be made productive, complexity needs to be stripped out for new builders. But these kinds of changes are extremely difficult to push through a democratic process, where consensus is slow and likely to be out of reach. This moment calls for a clear chain of command, and the right person to make the hard calls while moving fast.
This network exists because it's the brainchild of Const. There's no one better fit to command this era than him, and he’s already gotten to work on course correcting.
What has always been decentralized, and will stay that way, is the economic layer of Bittensor. Neither TAO nor subnet tokens were pre-mined. Both were earned through useful work, which over time distributed ownership outward, built one of the largest communities in crypto, and pulled in the most sophisticated talent in the field.
And participating in Bittensor has always been decentralized. Anyone can mine, validate, or launch a subnet without any permission. That's not changing. The centralization is around who steers the protocol.
My critique these past months has been the lack of process and communication from the core team. Const has said publicly that communication will improve, and I'm glad to see it. On process, he's said not yet. And I understand the instinct, because process is not what this moment calls for.
In the absence of process, we’ll still have coordination. Coordination means leveraging the army of independent contributors Bittensor has built, its single greatest asset. With the core team stating the purpose of an upgrade explicitly and signaling the timing of release, that army can provide feedback and conduct analysis before an upgrade ships. Plenty of people have built businesses on Bittensor, and coordination ensures they don’t end up operating blind.
The recent shorting proposal shows how this can work: the core team put out a writeup describing the upgrade and its purpose, alongside a simulated environment for the community to test the feature, try to break it, and give feedback before it goes live.
Eras, by definition, are meant to end. The era of partial centralization will end sooner than people think, and we'll cross into the era of decentralizing. That has always been the end state of Bittensor.
No crypto-AI project has accomplished more than Bittensor. Not in open-source breakthroughs, the scale of compute it coordinates, the agents and humans it pits against each other to produce useful work, or capital put to work. There’s nothing that comes close.
That is why Bittensor is worth fighting for. And it's why, through all the drama, late night upgrades, and changes, I stay convicted.
