Replies to yesterday's post seem to fall into three broad categories, so I'll cover them together:
Bittensor is about to delete a 25-year tech veteran's subnet, while he's still working on it.
Not for failing, just for price. Popularity.
Meanwhile the rules deciding which subnets get paid were rewritten three times in eight weeks.
Oh, and teams who paid $5M for subnet slots were told they'd always been mispriced. No refunds.
I've been invested in Bittensor for 18 months so I wrote down why I'm still here, and what I think needs to change: markcreaser.com/writing/chan…
Aug 5, 2026 · 11:56 AM UTC
9
2
43
7,264
The market has spoken.
It has. Babelbit has been judged by TAO holders. But Bittensor ISN'T the market. A translation subnet lives or dies on whether translation customers buy it, and no enterprise customer is checking the dereg queue on Taostats before they sign a contract.
I think Bittensor has created a problem where the TAO-holders are the electorate, so teams are forced to market inwards at the expense of customer acquisition.
In 2018 Tesla was the most shorted stock in the US and every analyst and his dog predicted bankruptcy. Elon ignored the share price, slept on the factory floor, and made Tesla the most valuable carmaker in the world two years later. In '97 Apple was nearly insolvent, and Michael Dell's advice was to shut it down and return capital to shareholders. Then came the iMac, the iPod, the iPhone.
I fear Bittensor would have lost both to dereg.
10
23
7,652
Where's the revenue?
I know how important revenue is in a business, but we need to be realistic. It often takes good companies years to generate revenue, and the AI mega corps like OpenAI and Anthropic still wouldn't be able to fund buybacks today.
I think OpenAI is still lossmaking, Anthropic might have just had its first profitable quarter. Both are fortunate to be backed by patient capital, and their backers cannot delete the company when the share price has a bad quarter.
And in most cases buybacks won't come from revenue, they'll come from SURPLUS cash, which is rarer still.
Yes, Bittensor buybacks aren't capital-return: alpha price is effectively the payroll, propping it is a pay-rise that attracts better miners. True. But that means you need to fund your payroll from revenue in year one or lose your best people. Even OpenAI can't do that. I don't know who can.
And yes, emissions are Bittensor's version of patient capital. But VC money never has to pass a daily market test in the same way that emissions do. Price falls, emissions fall, payroll thins, output degrades, the price falls further. It's a death spiral by design.
So many companies we hold up as examples of doing it right spent years as ugly ducklings, unpopular and burning other people's money. The difference is that nobody had the power to delete Amazon for a couple of slow quarters.
And while we all want subnet owners to go into the world and find revenue, look at what else we give them to think about that their VC-backed competitor doesn't: alpha price, emissions, rule changes, chain upgrades, staying active on X, posting in Telegram groups, answering questions in Discord.
Running a subnet is like trying to keep a Tamagotchi alive.
5
21
1,305
If you believe in it, buy the token yourself.
I could, but that'd be a whip-round, not a structural fix.
One buyer saving a subnet just throws another under the axe, and right now the next team in line is unlikely to be the one adding the least value to Bittensor.
And to be clear about what I'm NOT saying: I'm not saying we should gift emissions to subnets that don't deserve them. Let bad subnets have terrible token prices, exactly like the real world:
• Amazon fell 94% to under $6 in 2001
• AMD hit $1.61 in 2015, bankruptcy openly discussed
• Meta was down 76% in 2022
• Tesla was hours from insolvency on Christmas Eve 2008
Brutal price verdicts, all deserved. But Amazon kept its warehouses. AMD kept its engineers. The market punishes price but doesn't confiscate your business. Out there you only get deleted for running out of cash or fraud, something you DID, not something the chart did.
Bittensor’s scoreboard is also the executioner. And an executioner that selects on price doesn't take the parked slots and ghost teams first, it takes whoever's cheapest that week.
I'm not against the changes, or against dereg for slots that are genuinely abandoned.
I'm against a mechanism that doesn't seem to be able to discern an empty shell from a team that's actually trying.
2
24
952








