AI Cannot Solve QWERTY Problems
Better solutions don't always win
Happy Sunday and welcome to Investing in AI. I’m Rob May, CEO at Nova. We teach machines to understand brands, and make BrandGuard, which is a brand safety tool for AI (I don’t charge for this newsletter so, if you want to help me out, please send me a Fortune 1000 customer for BrandGuard). I also run the AI Innovator’s Community, a group in Boston and New York that cuts across all other affiliations and brings people together just focused on AI topics. We are launching groups in other cities soon, so stay tuned. And we have a great podcast if you are interested in that. In our last episode, Parasvil Patel from Radical Ventures gave his views on my “5 Contrarian AI Theses” post. Check it out.
Today I want to talk about QWERTY problems. I’m talking about the history of the keyboard most of us use. History says that QWERTY is a keyboard layout that is one of the slowest possible ways to type, and that it was designed this way because typing too fast on older typewriters could cause the letters to jam if two were typed two closely together. (Those of you over 40 may remember this problem). Regardless of whether or not that fact is true, it’s definitely slower than a DVORAK keyboard on which all the world typing speed records have been set. Why haven’t we all moved over?
Adoption is a tricky thing, particularly when it comes to standards. There is a book that isn’t that popular but worth a read called Network Power that discusses how standards get set, and what they mean in terms of “power” dynamics. From an idealistic perspective you don’t have to necessarily adopt a standard. You can do your own thing. But the form of power that standards create usually makes non-adoption impractical. This network power is part of the reason adoption of new things doesn’t happen. The value of the network is just more valuable than the better way of doing the thing.
What does that mean for AI tools? Many of them require new ways of working. They require new value chain steps like data labeling, new mindsets like thinking of outputs probabilistically, new quality and control workflows to deal with the scale at which AI can make things happen, and more. The more entrenched and universal the standard, the harder it is to change.
From an investing perspective, this creates an interesting dilemma. On the one hand, you want to evaluate a startup’s path to adoption and you prefer that it be an easy one. But on the other, if long term profitability tends to come from business model defensibility, well, there are business models people don’t attempt because of this QWERTY problem. Taking them on, and trying the very difficult path of changing a standard, could lead to bigger better outcomes for those companies.
When I consider investing in AI startups that have the QWERTY problem, here are the questions I ask myself.
Is the established path dominant enough that it will discourage most entrepreneurs? If so, I like that. Less competition, although admittedly a lower chance of success.
How strong is the value prop of the new way of doing things? This has to be looked at through an individual adoption lens, not a TAM lens. I could easily find some stats on how much more productive a DVORAK keyboard makes people, and how much keyboard typing happens around the world in a year and it’s probably a TAM worth hundreds of billions in productivity. But that’s really a meaningless number.
How much have people tried to change the standard? There are some areas, like payments, where people try frequently to change things. I prefer areas no one has looked at for a while because very often, the way to change it hasn’t been obvious simply because it’s been 15 years since someone has thought to seriously try.
And finally, can the team pivot around a bit? To take on these types of challenges, you need very flexible thinking. I want to see a team that will find a creative adoption path - preferably one that isn’t obvious from a simple market analysis but one you only find from living in the market for a while.
I bring up this topic because AI is moving VERY FAST. And as often happens in these speedy markets, thorough analysis goes by the wayside. Stop and ask yourself about adoption paths, and how AI might take on standards based businesses, when you are looking at those opportunities.
Thanks for reading.
Rob
Excellent analysis, Rob. I have been grappling with this exact dilemma myself both as a corporate innovation lead, advisor to AI startups and angel investor. Many Thanks for sharing.