Commoditize Your Complements

and why I think Llama is free

To the man who coined the phrase, “nothing in life is free”… have you been on GitHub lately?

Open-source is software that anyone can freely view, use, modify, and share because its code is publicly available on sites like Github and Huggingface. My last coding project alone was all open-source; I used the most powerful LLM model (Llama 3), in the most popular scripting language (Python), with the largest font library (Google Fonts), on the most powerful code editor (VS Code)- all without paying a dime or signing up for an account. What’s even more striking is that Meta, Google, and Microsoft are all key contributors behind these tools, respectively¹. The practice of big companies leading open source projects is actually quite common.

So, what’s their motive? Are they stealing my ideas or mining my data? Nope. These projects usually run locally on your computer (no internet needed) and are kept in check by the community, making sure there’s no foul play. The truth is these companies see their open-source projects as complements to their core business model. The goal is commoditize one’s complements.

Joel Spolsky is the co-founder of Fog Creek Software (creators of Trello and Glitch) and Stack Overflow. In 2002, he wrote his famous Strategy Letter V, which essentially predicts the meteoric rise of open-source software. Joel begins the essay by defining the economic terms of substitutes and complements:

Every product in the marketplace has substitutes and complements. A substitute is another product you might buy if the first product is too expensive. Chicken is a substitute for beef.

A complement is a product that you usually buy together with another product. Gas and cars are complements. Computer hardware is a classic complement of computer operating systems. And babysitters are a complement of dinner at fine restaurants.

I encourage you to try this exercise with your favorite product, business, or political party!

Spolsky explains that demand for a product increases as the price of its complements decreases. For example, a hotel owner in Miami should brace for more business if flights to Miami drop in price. At this point, Spolsky makes his definitive argument in bold:

Smart companies try to commoditize their products’ complements.

To put more bluntly, what needs to become $0 for customers to give us lots of money?

Spolsky cites several examples from the 90s that would make any Chicago Bulls fan nostalgic. The most relatable is Netscape’s decision to commodify its web browser. They could have charged for the browser, but they saw more money in the B2B server market. If everyday people could access a richer Internet with images and links for $0, demand for web companies would skyrocket. Those companies needed servers, and Netscape provided them. Next thing you know, ba-da-bing, ba-da-boom—Netscape’s founders became some of the wealthiest, most powerful men in Silicon Valley.

Commoditizing complements is still relevant today. Llama, Meta’s AI text generation competitor to OpenAI’s GPT and Google’s Gemini, is a possible example. Meta could have charged $10/month for AI access, but Zuck saw more profit in commodifying content. Meta owns Facebook and Instagram, the world’s largest content platforms, both fueled by a network effect of content creation and engagement. If people can freely create and consume content, engagement increases, drawing in advertisers and boosting revenue, which funds better AI, leading to even more creation. The flood of AI-generated content on Facebook and Instagram suggests this strategy will pay off, as the market will eventually balance with quality content. As long as content keeps flowing, Meta wins—while OpenAI and Google’s² margins take the hit.

Kibu & Commodification

Commoditizing complements seems to fit well with tech giants, as they have the time and resources to let their open-source projects grow. However, it’s never too early to plan! Our startup, Kibu, offers exercise & life skills content and administrative tooling to Disability Providers. Here’s how we might commoditize our complements:

  • Disability Providers & Staff Members: The more organizations and professionals in the disability space, the larger our total addressable market (TAM). Kibu can simplify the process of starting and managing a provider, similar to how Airbnb is commodifying hotels and Wonderschool is commodifying childcare.
  • Regulation: Tight regulations actually complement Kibu’s compliance tracking. We should commodify the communication of these regulations, making it easy for providers to see, understand, and care about the law. By engaging with consultants, influencers, and bloggers who excel in regulation, we can promote free knowledge and capitalize on offering compliance solutions.
  • Telehealth: Platforms that enable remote communication between providers and clients complement Kibu, as those sessions need tracking within our system. By offering free video communication solutions, we can create a flywheel effect: expanding access to disability care increases global billable hours, which boosts the TAM for the entire industry (especially us).
  • Streaming Media Devices: Affordable and plentiful streaming devices that connect to TVs and projectors complement Kibu’s content offerings. The more physical devices our customers can use to access Kibu, the lower the friction to purchase.
  • Educational Content: Our customers are incentivized to teach their clients life skills, job training, and independent living. We should aggregate all the free educational content available online on our platform. By curating links to games, printouts, PowerPoints, etc., we save our customers time while driving traffic to content creators. This not only adds value for our users but also encourages the market to create more content.

Footnotes

¹ While Meta is not the original creator of Python, they have contributed several updates to it to maintain its supremacy.

² Google is really its own economy: although closed-sourced to the public, other Google departments (Ads, YouTube, Pixel) will use Google AI as a complement to their product. But yeah, OpenAI and Anthropic are in tough situations.