mikecontango | τ,τ
@mikecontango
Followers
5K
Following
5K
Media
321
Statuses
4K
Managing Partner, @ContangoDigital | Co-Founder, @gTAOventures | Investor | Builder | Believer in Decentralization
Toronto
Joined January 2020
Agree with this statement whole heartedly. Bittensor is built to evolve, and must evolve to increase efficiency and maximize productivity. A non-evolving intelligence network is an oxymoron. $TAO
@here4impact Bittensor isn’t anything like a football game. We’re experimenting with a completely new field called incentive computing. You’re not going to get that right on the first try, or the second, or the third. If the expectation is that all the kinks should be worked out by now,
1
1
14
Appreciate the time today @gordonfrayne. Thank you for everything you do for the space!🙏
Sat down with @mikecontango and @0xVonBismarck from @ProjectRubicon this morning to unpack how Bittensor connects to EVM, and what unlocking real liquidity means for subnets. Here's what we covered 👇 03:00 — How Mike and Victor went from TradFi / VC into Bittensor $TAO 07:30
3
0
10
Learn more ↓ https://t.co/z56Rt9zw96
0
0
1
Trading #Bittensor subents on @base just got easier. xAlpha tokens from @Rubicon_Bridge are now available directly in the @subwalletapp app! $TAO
xAlpha tokens are now integrated on @base & Subtensor EVM ✨ SubWallet now supports liquid-staked Alpha assets such as xTAO, xSN4, xSN8 and more as they expand across Base (through @Rubicon_Bridge by @gtaoventures & Chainlink CCIP), and across Subtensor EVM You can view
0
4
29
Last week, @Base unveiled the Base-Solana Bridge, secured by Chainlink CCIP. Now, @Coinbase officially announced is has adopted @Chainlink CCIP as the exclusive bridging solution for all Coinbase Wrapped Assets. More proof @chainlink CCIP is industry standard for cross chain
5
3
37
14/ In one sentence: Architects build the weapons. Adversaries wage the attack. Oracles sell the battle externally. Alpha coordinates, rewards, buys itself back, and compounds the entire loop. $TAO @trishoolai
0
2
8
13/ Long-term, SN23 converts alignment stress-testing into a competitive marketplace: • Incentivizing attack research • Incentivizing modular infrastructure • Benchmarking AI safety with economic pressure behind it The subnet rewards the creation of better tools, better
1
0
5
12/ Trishool’s core design principle: Alpha is not just a reward token — it is an essential production input. It must be: • Burned to use components • Staked to earn yield • Held to receive revenue share • Bought back by the protocol itself via Oracle revenue This creates
1
0
5
11/ Alpha holders earn from two sources: • Validator staking yield • 15% share of real Oracle revenue Alpha becomes: • A productive asset • A governance and coordination primitive • A direct claim on subnet alignment demand
1
0
6
10/ Validators coordinate and secure the entire system. They: • Enforce correct evaluations • Prevent manipulation • Distribute rewards fairly Validators earn 41% of Alpha emissions, and 18% flows to the subnet owner for maintenance and upgrades.
1
0
5
9/ Top Adversaries benefit twice: • They earn performance-based Alpha emissions • They receive 15% of external client revenue As revenues grow, leading miners are incentivized to use more components, which means burning more Alpha. Success amplifies Alpha consumption.
1
0
4
8/ Clients pay Oracles to benchmark their models. That revenue is distributed back into the system: • 30% → Alpha buyback (ψ) and burn • 15% → Alpha holders • 15% → Adversaries • 40% → Subnet Owner This ties real-world demand directly into token flow, and introduces an
1
0
5
7/ Oracles form the execution and verification layer. They: • Run the best agents • Against real frontier models • For real enterprise clients • Under standardized evaluation conditions This is where external revenue enters the subnet.
1
0
4
6/ This creates two pressure loops driving continuous improvement: • Architects compete for component adoption + performance efficiency • Adversaries compete for attack success + Alpha efficiency Bad components die economically. Bad agents die competitively.
1
1
6
5/ Adversaries assemble the most capable agents using these components. Their objective is simple: Jailbreak frontier models better than anyone else. They are rewarded in Alpha based on performance, not volume.
1
0
6
4/ To use a component, Adversaries must burn Alpha (ψ). This burn is economically recycled back to Architects, who are rewarded in Alpha based purely on usage. This forces: • Architects → to ship performant, cost-efficient building blocks • Adversaries → to optimize agent
1
1
6
3/ Architects build the subnet’s modular component layer: prompting tools, attack primitives, routing logic, memory modules, and more. These components form a shared library that Adversaries can plug into their agents.
1
0
6
2/ There are three miner roles in SN23: • Architects – build modular infrastructure components • Adversaries – build jailbreak agents • Oracles – evaluate agents against real client models Each role earns Alpha in a way tied directly to economic utility, not vanity metrics.
1
0
6
1/ Trishool (SN23) is an #AI-alignment benchmarking subnet on Bittensor that incentivizes the creation of the best jailbreaking agents to stress-test frontier models. Its tokenomics tightly align infrastructure builders, agent developers, validators, and capital providers into a
1
0
6