TIG: The Snowball Effect

tigfoundation avatar
The Innovation Game (𝔦, 𝔦)
@tigfoundationSep 20, 2025

The Spaces brought together John and a host to unpack how TIG is moving from early traction to a durable network effect, why its model scales beyond traditional labs, and what that implies for AI progress and value capture. They recapped John’s widely shared interview that distilled the “TIG pill,” then detailed the current cadence (6 live challenges, 8 in development) and why onboarding is temporarily capped while documentation and legal scaffolding mature. They explained the migration of liquidity to Aerodrome for stronger incentives and Base alignment, plus complementary dialogues with ResearchHub (co-founded by Brian Armstrong). On tokens, John outlined the founders’ vesting cliff, reiterated they don’t plan to sell, and noted the upside of higher circulating market cap for listings. A core segment assessed AGI timelines: John projects ~10 years, citing the need for algorithmic advances beyond transformers, hybrid LLM + program search systems, and a “mother of all datasets” that codifies inventive know‑how and strategies to get “unstuck.” TIG’s design channels this: researchers (often non‑crypto) self-select into challenges, mentor newcomers, and generate state‑of‑the‑art algorithms that will compound into weekly breakthroughs by 2026. TIG will capture value via licensing of an unforkable IP portfolio. The team is adding a researcher to accelerate challenge design, outreach, and enterprise BD. They closed by emphasizing the snowball effect: no incumbent pushback, expanding into a vacuum, and an open, self‑policing scientific community.

Tig Space: Snowball Effect, AGI Timeline, Liquidity, and Scaling Science

Participants and Roles

  • John: Scientist/founder leading Tig; primary technical voice on challenges, AGI, protocol design, and strategy.
  • Host (Speaker 2): Space moderator and community voice; VC background; articulates investor/community questions and thesis around Tig’s network effects and value capture.
  • Others mentioned:
    • Ying (Tig team; challenges pipeline coordination).
    • Ian (Baseline Podcast host who interviewed John).
    • Brian Armstrong (Coinbase/Base; co-founded Research Hub).
    • Research Hub (aligned DeSci project; complementary to Tig).
    • Tibo Vidal (state-of-the-art vehicle routing expert engaged via Tig).
    • Francois Charlie (Google; talk on LLM+symbolic program search).

Context and Momentum: Interview, Decentralization, and Why Science

  • John’s interview on the Baseline Podcast (hosted by Ian) became the channel’s most-viewed episode (now ~3x the next-closest). Both credit the quality of the interviewer’s preparation and probing questions for clearly conveying the “Tig pill”: what makes Tig different and valuable.
  • Decentralization for science: The Host, no longer a “decentralization maxi,” still argues science is among the domains most likely to benefit from decentralization due to constraints in centralized systems (governments, institutions). John adds Tig’s distinctive advantage is not necessarily early speed but the absence of upper scaling limits that paralyze traditional organizations (hiring, cost of uncertain R&D). He likens Tig’s steady, unstoppable growth to a “Terminator” that “never stops.”

Challenges Pipeline: Where We Are and Why It’s Accelerating

  • Current state: 6 live challenges and 8 in development (14 total). The Host was surprised by the pace given the team had said they were “holding back.”
  • Why “holding back”: Documentation and legal frameworks for challenge owners/designers took longer than expected; once ready, the eight eager creators started simultaneously, likely finishing around the same time (a benign bottleneck). The team has paused onboarding new challenge builds until this batch is through.
  • Modularity and permissionlessness:
    • Contributors can pick their favorite domains and move between them (e.g., vector search one day, vehicle routing the next).
    • Challenge owners work independently; outputs slot into Tig without central coordination bottlenecks.
    • Documentation is improving continuously; challenge owners now help onboard and advise newcomers, reducing load on the core team.
  • Diversity as a design principle: Tig targets biology, robotics, AI, and beyond—ultimately hundreds of challenges across scientific domains. This breadth is feasible because of modular design and pay-for-success mechanics.

Network Effects: Snowballing via Referrals and Reputation

  • Early challenge creators are top scientists; many are not “crypto people.” Their referrals have driven the new pipeline (“quietly taking shape”).
  • Community self-policing like open source: Strong norms, high standards, and top talent set a precedent newcomers emulate.
  • The Host’s thesis: As Tig’s outputs become too compelling to ignore (e.g., frequent state-of-the-art improvements), external awareness will explode. When the broader research community and media notice, expect a “Mongol horde” of researchers to arrive—on both innovation and challenge creation.

Breakthrough Cadence: From Months to Weeks as Challenges Scale

  • Achievements to date:
    • Knapsack challenge already improved to new SOTA (with further gains likely soon).
    • Forum feedback suggests Tig’s vector search approach may yield up to 100x improvements in certain edge cases.
    • Tibo Vidal (incumbent SOTA in vehicle routing) is working through Tig, underscoring expert confidence and alignment.
  • Scaling projection:
    • With ~5–6 challenges now, cadence is roughly one breakthrough every 2–3 months.
    • With ~50 challenges (plausible by end of 2026), cadence could approach weekly SOTA-level results.
    • Pay-on-success economics eliminate the need to employ researchers; productivity scales approximately linearly with number of challenges. Over time, Tig could eclipse any single lab’s output, then scale to hundreds/thousands of challenges across domains.
  • Value capture: Tig aims to license state-of-the-art, licensable algorithms (e.g., hypergraph partitioning, optimizers). If an algorithm is 100x more efficient, competitors are compelled to adopt it, driving licensing demand.

Liquidity and Ecosystem Alignment (Base, Aerodrome)

  • LP migration: Uniswap liquidity unlocked and was migrated to Aerodrome (Base’s native DEX), which offers more attractive incentives.
  • Strategic alignment: Aerodrome aligns Tig with Base/Coinbase. Tig is not backed by Coinbase Ventures (at least not yet), but Base has amplified Tig’s announcements. This alignment benefits Tig’s positioning within the Base ecosystem.

Coinbase, Research Hub, and DeSci Alignment

  • Relationship vectors: John confirms Tig has a positive relationship with Base and has been speaking with Research Hub (co-founded by Brian Armstrong).
  • Complementarity: Tig and Research Hub cover different layers of the DeSci stack; they are not competitors. Coinbase/Research Hub share genuine interest in science.
  • Why Base: One factor was Brian Armstrong’s public commitment to donate a portion of his income to scientific research—an alignment with Tig’s research-centric mission.

Founders’ Tokens, Cliff, and Selling Stance

  • Vesting: 1-year lock completed/imminent; followed by 2 years linear vesting.
  • John’s stance:
    • No intention to sell founder tokens; emphasizes long-term commitment.
    • Tig was self-funded for ~2–2.5 years to launch the protocol and token, avoiding early VC warrant deals at “pennies on the dollar.”
    • John was originally against locks (preferring the stronger signal of choosing not to sell when one could), but locks were implemented due to investor pressure at the time. Practically, it made no difference because founders weren’t going to sell anyway.
  • Market-cap optics: Locked tokens don’t count as circulating supply; unlocking can lift reported market cap, which may help centralized exchange listing prospects. No plan to re-lock.
  • Liquidity additions: Additional liquidity is under discussion; if added, it would be matched with stables to avoid one-sided sell pressure.

AGI Timing and the “Mother of All Datasets”

  • John’s estimate: Not within 2 years; at least 10 years. He ballparks 2036 for AGI-like capability (e.g., reliable memory over days/weeks, very low hallucinations, outperforming an intern on complex tasks).
  • Why not sooner:
    • Current LLMs are energy-inefficient versus the brain (~15W) and have hit a scaling plateau: GPT-4-level scaling stalled, and subsequent gains appear to be “tune-ups,” not fundamental architectural leaps.
    • Progress now requires algorithmic breakthroughs beyond transformers and better optimizers.
  • Mother of all datasets:
    • AGI needs “inventive know-how”—the tacit strategies researchers use to get unstuck. These are barely codified in literature but are codifiable.
    • Training AIs on merely the clean progression of technical literature omits the essential “unsticking” heuristics that real discovery requires.
    • Therefore, to invent the next algorithms, AIs must be trained on codified inventive know-how—creating the “mother of all datasets.”
  • LLM + Symbolic Program Search:
    • John cites a talk by Francois Charlie (Google) on coupling LLMs with analytical/symbolic reasoning modules (program search).
    • His thesis adds that LLMs themselves must be improved with inventive know-how data to drive real algorithm invention.
  • Near-term expectation: Chatbot-like applications may plateau while frontier AIs shift toward algorithm discovery (new architectures, optimizers), releasing the current bottleneck and triggering the next surge.

From Mining Hashes to Mining Algorithms: The Monetization Loop

  • John’s analogy:
    • Future AIs will “mine” for algorithms via compute/energy-intensive search (akin to ASICs).
    • Tig is like the Bitcoin network for algorithms: discoverers submit to Tig, instantly monetize through benchmark markets and licensing, and reinvest in compute.
    • All results remain open and composable in Tig, with “benchmarkers” ensuring the best current methods are known and built upon.

Business Development, Talent, and Organizational Design

  • Researchers as BD:
    • Tig Labs is adding another researcher. Researchers are best positioned to conduct outreach, qualify enterprise needs, and map problems to specific algorithms.
    • Technical fluency is essential in conversations with enterprises and academics; generic BD hires often lack the depth needed.
  • Talent quality: Existing researchers have “spectacular” productivity and strong soft skills, aiding outreach and adoption.

Upcoming and Active Challenge Areas

  • In development or on the radar: Robotics; zk process; hyperparameter optimization; computational fluid dynamics; hypergraph partitioning; optimizer (neural network training); vector search; vehicle routing; knapsack (already improved).
  • Strategy: Maintain diversity with a slight bias toward AI because improvements there compound into other domains (medicines, fluid dynamics, etc.). Avoid multi-year formulations where possible; the optimizer challenge was a rare “marathon” that proved worth the push.

Risks, “FUD,” and Why the Team Sees an Open Field

  • Host’s continual “FUD yourself” practice can’t identify meaningful blockers:
    • No incumbent being disrupted in a way that provokes a counterforce; Tig expands into a “vacuum.”
    • Governments are unlikely to resist more efficient, market-driven allocation; savings can be repurposed elsewhere.
    • Strong network effects via scientists and an unforkable IP portfolio make it very hard to copy Tig’s accumulated advantage.
    • By the time mass awareness arrives, Tig’s head start could be decisive.
  • Monopolies for public goods: Host quips that monopolies can be good if they produce a public good transparently, in open source. John agrees.

Community and Miscellany

  • Housekeeping topics addressed: liquidity migration; founder vesting; potential additional LP; reassurance about contributor incentives (serious scientists are aligned long-term rather than extractive).
  • Weekly community touchpoint: Spaces continue weekly; occasional token lotteries for listeners.

Key Takeaways

  • Tig’s growth is transitioning from carefully guided to documentation-driven, enabling many more high-quality, independent challenge owners.
  • Referral-based onboarding of top scientists is already producing state-of-the-art results; with 50+ challenges, weekly breakthroughs seem plausible.
  • Tig’s pay-for-success model and modularity eliminate central bottlenecks and make scaling across domains feasible.
  • Strategic alignment with Base (Aerodrome LP, amplification), conversations with Research Hub, and a non-extractive founder stance bolster credibility.
  • AGI is likely a decade away; the path runs through algorithmic innovation powered by codified inventive know-how and LLM+symbolic program search—areas where Tig’s algorithm engine and licensing can become central.
  • As AIs “mine algorithms,” Tig provides the open, liquid, value-capturing venue to submit, benchmark, and monetize them—closing the loop that funds further discovery.