The “libre” paradox

There is a great amount of discordance in the worldwide community at large regarding what kinds of software should be made free, open-source, or commercial. Even I, who am not a developer of any prominent software, have had to tackle this question myself, especially after the Aseprite fiasco regarding its conversion from commercial GPLv2 to commercial closed-source.

My empirical findings about software production models is that while commercial software can achieve results quickly and efficiently, open-source software runs on ideas and thus tend to achieve results with greater quality. Developers might be hired to write a specific program in six months, yet a developer has all the time in the world to think about the design of a personal project before even putting down a line of code. Moreover, academics (assuming, of course, that academics are the ones who work on FOSS projects, since they are too busy for a full-time job, but are keen to write code for societal good) have an affinity for peer review, encouraging only the best development and security practices, under risk of scrutiny otherwise.

It is of no surprise, then, why companies tend to cherry-pick code and design from FOSS projects to fuel something slightly better.

When a new idea is introduced for the first time, it is competition and money that drive results. When Bell Labs et al. dominated computing research for a long time, and the threatening Soviet Union begrudged the United States government to fund the research of NASA, these are prime examples of manifestations of driving factors for research and innovation.

But neither Bell Labs and NASA ever sold products to consumers. Instead, other companies were founded to fill this gap – not to create something radically new (often, when this occurs, they miserably fail or dramatically succeed), but rather to simply take the next step. The research has already been complete – just put it in a box, along with an instructions manual, and sell it to consumers. There’s nothing like it on the market, so it’s a perfect choice for consumers. Rake in the cash. Corner the market. And soon, a new company will form itself to take yet another baby step in innovation, and that one will be fruitful too.

When the innovation has become so clear and obvious to the public that it can be learned by undergraduates or any interested student, it is then time to charitably introduce the innovation to others. The modern computer has existed for a long time, yet Eben Upton and the Raspberry Pi Foundation took the “small” step of putting a SoC on a small board and selling it for $35. At the time, I don’t think it would have been easy to find a technologically up-to-date, general-purpose computing device at that price point and form factor. But because the Raspberry Pi Foundation did it, now many businesses exist for the sole purpose of manufacturing and selling low-cost single-board computers. As a result of this work of charity, computers are now easily accessible to all. What’s more, students can and must take courses covering the architecture of the modern computer, and some students are even tasked with constructing one from scratch.

Likewise, once an open-source project is done on a particular topic, that particular topic is essentially “done“. There are not many businesses out there that sell consumer operating systems anymore; if people seek a free operating system, there’s GNU. It’s done; why look further? Any improvements needed are a code contribution away, solving the problem for thousands of others as well. Why should companies strive to produce new modeling software if they must compete with programs like Blender and existing commercial software such as Maya?

My observation is that open-source software is the endgame. It is impossible for a commercial software with the same features as an open-source program to compete with each other; the open-source program will win consistently. Conversely, commercial software stems from open-source algorithms waiting to be applied, be it TensorFlow or Opus.

Basically, it makes sense to start a company to churn out commercial software if one is willing to apply existing research to consumer applications (take small steps); join a larger company to rapidly develop and deploy something innovative; or join academia to write about theory in its most idealistic form.

Under these observations, startup businesses fail because they attempt to innovate too much too quickly. The job is not to innovate immensely all at once – the job is to found a business under a basic, yet promising idea (the seed), produce results, and then continue taking small, gradual steps toward innovation. The rate of innovation will be unquestionable by investors – if you survive for two years, putting your new features and products at a healthy pace, then people will naturally predict the same rate for the coming future, and be more willing to invest.

Yet you would never find enough resources to make a successful startup for, say, building giant mechs or launching payloads into space. There’s just too much research to be done, and the many people who are capable (and in demand) to perform this research need coin to sustain themselves. In contrast, the military can pour any amount of money they wish to a particular project, and they could have a big walking mech that looks like the one from Avatar in less than 36 months. (I’d wager the military has already been working on this as a top-secret project.)

But do you see how much we have departed from the idea of “libre?” My conclusion is this: businesses do things quickly, while charitable people do things right. Once the research has been completed and the applications have been pitched and sold, it is then time to transition and spread the innovation to the general public. That is the cycle of innovation.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.