The Innovation Game

Crypto’s strength lies in its ability to enable coordination at scale. By setting a shared objective, crypto incentivizes stakeholders to push that objective forward. This was Bitcoin’s breakthrough: all participants act in their self-interest, resulting in the largest global supercomputer and peer-to-peer money network. The same concept is applied to other networks that utilize useful proof of work, meaning that through competition, participants push forward an exogenous objective.

One such network is The Innovation Game (TIG), which operates like a global, open-source research lab focused on algorithm optimization. Algorithms solve a wide range of problems, so improvements in specific categories can have enormous implications for business and science. TIG focuses on asymmetric problems: easy to verify, difficult to compute. This approach is critical because it minimizes the attack surface, making it easier to validate quality work.

How TIG’s Research Lab Works

In TIG’s decentralized research lab, innovators create algorithms to solve specific problems and miners (called benchmarkers) test for the most efficient algorithms. Miners are incentivized to complete problems as efficiently as possible, naturally gravitating toward the best algorithms. This dynamic allows miners to act as decentralized evaluators of the algorithms.

TIG represents a market that has remained largely untapped outside academic and private institutions. The result of competition among innovators and benchmarkers is a catalog of state-of-the-art algorithms for solving common optimization problems. This intellectual property (IP) can be licensed to companies like Google, Microsoft, and OpenAI, where even a slight improvement in an algorithm can result in billions of dollars in revenue or savings. At a high level, the flow looks like the diagram below. 

In practice, TIG has an intricate framework for managing the IP of stakeholders in the network. The distribution licenses ensure the TIG Foundation is rewarded for their IP, and the game licenses ensure that each participant is rewarded for their contributions to the TIG Foundation.

Game Licenses

1. The TIG Innovator Outbound Game Licence: allows innovators to build upon previously submitted algorithms.
2. The TIG Benchmarker Outbound Game Licence: allows benchmarkers (miners) to leverage optimizations other miners have open-sourced to the network. 
3. The TIG Inbound Game Licence: allows the TIG Foundation to secure IP rights from the innovations on the network. 

Innovation Distribution Licenses

1. The TIG Open Data Licence: allows free use of algorithms on the TIG network with the contingency that improvements are shared back with the TIG network and any commercialization is open-sourced. 
2. The TIG Commercial Licence: allows closed use of TIG IP for a licensing fee. This is how revenue will accrue to the TIG Foundation. 

The Market for Algorithm Improvements

At first glance, TIG might seem abstract: aren’t most well-known problems already solved? Is there a market for algorithm improvements? These problems have solutions, but they can always be optimized to increase speed and efficiency or be adapted to run on faster hardware like GPUs. Currently, most of these refinements happen in academia or within private companies. TIG bridges the gap by enabling researchers to be paid for their work on merit, with no strings attached.

The problems tackled in the TIG network have broad applications. One example is gradient descent which is a foundational technique in machine learning, especially in training models such as linear regression, logistic regression, and neural networks. With the increasing importance of AI, improvements to gradient descent algorithms could vastly improve training speed, model performance, and energy efficiency. 

Global spending on AI was estimated to be over $150 billion in 2023, and that spending is increasing rapidly. A 10% improvement to gradient descent would be worth billions.  As an example, if we look at the annual AI compute costs of some of the most prominent players in the AI race, we can back into reasonable cost savings and the amount they would be willing to pay for state-of-the-art gradient descent. Meta is spending $20 billion in 2024 on AI infrastructure for Llama 4. OpenAI is rumored to be spending $2 billion on GPT5. A 10% improvement to gradient descent would result in tens of billions of savings over the course of a few years. The opportunity is enormous.

Another example is the Knapsack Problem, a classic optimization challenge in computer science and mathematics. It involves selecting items with given weights and values to maximize the total value without exceeding a weight capacity. This problem is relevant in areas like portfolio optimization, logistics, and cargo loading.

As an example, a small improvement in the algorithm for solving the knapsack problem could significantly impact logistics efficiency. In 2018, the logistics industry’s total costs, including transportation and warehousing, reached $1.6 trillion. Here are a few ways improvements could impact the industry:

• Optimized Load Planning: Improved algorithms can maximize space utilization and minimize costs in logistics load planning, enhancing transportation efficiency.
• Resource Allocation: More efficient algorithms can optimize the use of vehicles, storage, and labor, reducing waste.
• Cost Savings: Efficient load planning and resource allocation can significantly reduce fuel, labor, and storage costs for logistics companies.
• Increased Profitability: Enhanced efficiency and cost savings from optimized logistics operations boost profitability and market competitiveness for companies.

Creating Markets for Digital Commodities

Creating markets for digital commodities only works if you can verify their quality. To date, this has been a major challenge for DePIN networks. How do you evaluate the relative quality of two LLM inference responses? How do you ensure a benchmarking system can’t be gamed? Output evaluation is often more difficult than it appears.

TIG’s narrow focus on algorithm optimization allows for a smaller attack surface, increasing the network’s probability of success. If a rule set is too wide, it creates an adversarial environment between rule breakers and core developers who need to constantly mitigate edge cases. In TIG’s case, it instead creates an arms race among peers, ultimately improving the algorithms on the network. 

Its incentive structure is built to attract early talent through high inflation that tapers off as the network matures. TIG’s IP will become increasingly valuable as the protocol grows, generating licensing revenue that flows back to network stakeholders. Moreover, the decentralized nature of TIG makes it uniquely resilient. It’s a global marketplace, open to participation from anyone with merit and drive.

Algorithm optimization has never been approached in this decentralized, open-source manner, creating an entirely new asset class—digital commodities backed by verifiable IP. The TIG network bridges the gap between academia and industry, enabling innovators to get paid for their work without the restrictions of traditional institutions. 

TIG’s model is self-sustaining, and as the network gains traction, the value of its licensed algorithms will compound. The potential ROI is enormous, especially for early participants who help grow the network’s intellectual property catalog. Just as Bitcoin rewarded its early miners, TIG offers a similarly asymmetric opportunity.

Join the Game: Become a TIG Innovator or Miner

Whether you’re an innovator looking to test cutting-edge algorithms or a miner eager to benchmark and earn rewards, TIG offers a chance to contribute to the future of algorithm optimization while earning tangible incentives. Come compete!