Vibes won’t make your software successful

I regret to inform you that vibes don’t pay the bills. Specifically, open source vibes. For as long as I’ve followed open source, there has been a tendency for the industry to focus on normative values of open source, i.e., why people should embrace open source, as if it were a morally superior way to build software. But software, whether open source or proprietary, is just a tool to get things done. Whether open source is “good” and closed source is “bad” depends entirely on whether it helps developers and customers more effectively get stuff done.

As Redis CEO Rowan Trollope put it to me recently, for the average developer it’s not a question of open or closed licensing but rather, “Does this thing offer something unique and differentiated … that I need in my application?”

With that in mind, it’s worth paying attention to where the money is moving in AI, the hottest market in tech. The poster child for open source AI is Hugging Face, which generates a nice profit by positioning itself as ground zero for open source AI models. Indeed, the company goes further, touting open source as the only way to do AI correctly. Fine, except for this inconvenient truth on its site: “Llama is one of the first open source LLMs to have outperformed/matched closed source ones.” Many would say Llama isn’t open source at all but … details! Yet the real money in AI is going to the hyperscale clouds that are less idealistic about open source and a lot more focused on customer convenience.

In other words, open source matters a lot, but maybe not in the ways we sometimes think.

Follow the money

One of the biggest beneficiaries of open source is the company that for a long time seemed to be taking but not giving back to open source: AWS. The reality of AWS’s contributions to open source, to be fair, was always more nuanced than critics suggested. It’s true, the company has never used open source as a business strategy as overtly as, say, Google, but AWS’s primary focus was always on making open source easier for enterprises to consume.

During the past few years, under the leadership of David Nalley, AWS has managed to flip the industry narrative around itself and open source. AWS has become less didactic and more pragmatic about open source while maintaining an optimism that open source solutions tend to work better for customers. In other words, less waving of ideological banners and more touting of how customers benefit from AWS’s increased involvement in projects like OpenSearch, Postgres, and more. The more the company has worked to make open source better for customers, the more customers have responded by buying AWS services.

Another example is the Linux Foundation. Despite being funded by corporate contributions, which could arguably make the Linux Foundation more likely to placate corporate donors with weakened definitions of open source, the Linux Foundation has become the industry’s most stalwart standard bearer for uncompromising open source. This advocacy for open source rarely plays the same saccharine notes we hear from other open source nonprofits. The secret of the Linux Foundation’s success? Its ability to tie open source to business value. As Linux Foundation Executive Director Jim Zemlin argues, “Open source software has substantial economic value to companies and … its value will increase for most organizations the longer and more they use it.”

The value of open source

This is always at the heart of the Linux Foundation’s open source advocacy: Contributing to and consuming open source is good for business. By selling that message, the Linux Foundation also does more to ensure the sustainability of open source than its vibes-based peers (such as the Free Software Foundation), which historically have heavily emphasized ideological arguments for free and open source software, thereby limiting its appeal to any but the most philosophically committed.

The best open source advocacy, in short, tends to focus on how it helps developers and their employers more effectively get work done rather than on ideological arguments about the right way to build or consume software.