Google's Kodak Moment

The one thing that I had hoped that businesses would learn is that you shouldn't be put out of business by something that you invent. As a reference for readers not familiar with Kodak; Kodak invented the sensor for the digital camera, and digital cameras ended up killing Kodak by obviating the need for film, and though I never went to business school, I do know that Kodak is used as a case study of what not to do when your company invents something transformative. In much the same way, Google Research invented/created many of the key technologies that underpin ChatGPT. In terms of the basic technology, Google Research invented the Transformer Layer the key piece of technology for Large Language Models (LLMs), created most of the modern frameworks like TensorFlow to make working with neural networks “easy”, and even had a prototype of ChatGPT called Bard 2 years prior to the release of ChatGPT 3. I would even hypothesize to say that Bard might be technically more sophisticated than GPT4 (sans the period of training that made GPT4 so incredible), but like Kodak, they chose to suppress their version of Bard for the simple reason that they couldn’t figure out a way to monetize the technology, and they knew that it was pretty much something that could hurt one of their primary sources of revenue. Now I don’t have a crystal ball in whether or not Google will go the way of all the prior search engines companies like Yahoo, but if it does, their response to the use of large language models for information retrieval will be seen?as the canary in the coal mine that hinted at the downfall of google.

I don’t think that Microsoft is necessarily more visionary than Google has been in terms of their adoption of the technology. In much the same way that Google failed to exploit LLMs more widely was due to their inability to monetize it, Microsoft has a pretty clear path for these LLMs to show business value. Unlike Google whose business model is selling human attention, Microsoft’s business model relies on selling software products that help us create new things. Therefore these technologies that make it easier for us to perform creative tasks are things that not only slot neatly into their product portfolio, but also increase the stickiness of their existing products. ChatGPT embedded into a product like Word makes it better, makes it a more attractive solution than Google Docs or Pages. Github Copilot when embedded into an IDE makes the IDE and software development toolchain more attractive, both things that make Microsoft money. Therefore because it easily fits into their existing business model, Microsoft rapidly adopted the new technology. And as a side note, having used GPT4 enabled Bing, it still doesn’t move the needle to use Bing instead of Google Search only because it doesn’t significantly improve the experience.

The real reason stories like this make me sad is because it decreases the perceived value of corporate research. I feel that companies are less likely to invest in basic science research if they a) can’t find direct uses for the technologies and b) run the risk of being put out of business by something they discover. I’m still sad that Bell labs wasn’t able to commercialize things they invented such as C, Unix and the germanium transistor. In hindsight, these technologies are the underpinnings of our modern connected world, and had Bell been able to monetize these inventions, we probably would still have a large corporate lab churning out Nobel Prizes once every few years, and despite any of the missteps that Google might have had as a corporate entity, I feel that much of the research that they have done has been a net positive for society.?

While I don’t think that the CEOs are necessarily to blame for not having the vision to do the right thing and embrace transformative technology, I do feel that the existing conventional wisdom for what a corporation should look like does negatively impact a corporation’s ability to take advantage of their own research. For instance in the prior example of Bell Labs, they had the potential to be Microsoft, Intel, Apple and maybe even Google all rolled into one, had they had the capability of monetizing everything that came out of their labs. In the context of Google, their work in Alpha-Fold should have at least made them major competitors in the drug discovery space (my own field), instead they have paved the way for lots of small startups like Generate Biosciences of which they will have no upside if these companies succeed.

I find it somewhat ironic that while the West prides itself on capitalism, their corporations are run like centrally planned command economies. I think for companies to achieve the same level of success as the wider society, they have to embrace some of the messiness inherent in capitalism. Companies like Google and the former Bell labs ought to be actively convincing their employees to take notice of the innovations that come out of their labs and try to create a start-up that leverages that technology. In my ideal world the first thing that Google should have done after the publication of Alpha Fold 2, was to see whether some internal employee had a business plan for it, and provided that they did, provided some seed money to get that start-up idea off the ground. In much the same way universities like Stanford encourage professors to create their own startups, companies like Google ought to do the same with the understanding that if these start-ups succeed, these companies could either derive a steady revenue stream from these discoveries, or at least have an avenue by which to pivot to a new revenue stream. Given the amount of sheer money that these companies have made and banked in the last two decades, I believe that a better use of their money besides acquisitions, stock buybacks would have been to spin a couple of dozen startups out from their research labs founded by their own employees, because it’s not sufficient to experiment on technology, but it's also necessary to experiment on different business models that leverage the technology, and if these companies really have hired the best and the brightest of their fields, who better to make these advances than their own employees.?