No End of Computer History (Yet)

At USV we have been discussing a cooling internet universe for several years now. We have largely stayed away from backing strictly sales driven companies as they are exposed to competitors who can spend more money on sales and marketing. We have relatively early on branched out into investing in blockchain/crypto with the belief that permissionless data could unleash a new wave of innovation the way that permissionless publishing did with the web. A recent essay by John Luttig titled “When Tailwinds Vanish” lays out the case for cooling or maturity of the internet industry and provides some compelling data on decelerating growth rates.

And yet, underneath everything, there is still so much potential for the radically new and different. I have already mentioned blockchain/crypto which, despite setbacks and challenges of all sorts, continues to grow. But there is the beginning of another trend that I am extremely excited about: a reinvention of how we do things on computers through new user experiences enabled by machine learning.

We are already seeing some of what’s possible by innovating on user experience. Airtable is changing the way people are approaching databases. Notion is unifying different document types. Other interesting projects that are a bit earlier include Almanac and Roam Research.  I am aware of teams working on fundamentally new approaches to communication, to information storage/retrieval and to web browsing. While these are in their infancy today they have the potential to be transformative (not linking to these as they are all trying to fly under the radar a bit).

To understand what may be happening it is useful to go back into the past. To make computers intelligible and accessible to users, programmers and designers imported metaphors from the analog world: a desktop, files, folders, tabs, windows. Gradually some of these have lost in importance. For example, search has made it so that people no longer have a lot of need for complex well-maintained folder structures. More recently we have seen the rise of voice interfaces with the likes of Alexa which do away with all of this past baggage and aim for a conversational interaction instead.

And while voice is great, our ability to absorb and process complex information relies heavily on visualization, on the ability to go back and forth over text, as well as making connections between different pieces of information. Yes, Google Docs and Gmail have become smarter but they are adding machine learning to an existing design. The really big opportunity is to rethink the UI/UX for these types of applications from the ground up, leveraging everything that’s in place now including ubiquitous smartphones and high bandwidth connections.

I don’t profess to know what will ultimately succeed. Lots of experiments will fail to go anywhere. But I have no doubt that it can and will be done. There is just so much knowledge, structure and connection in the text and images that we process every day that is suddenly accessible to machines but not yet at our fingertips. For example, I am writing this in Google Docs and while I know that I have written about closely related topics before, I am getting zero assistance in making these connections (or relevant content from others for that matter).

Bringing it back to Luttig’s post and the question of cooling. Is it realistic to expect exponential product led growth in an era of displacement? The hurdle for displacement is obviously much higher than for initial adoption. Yet the history of technology is full of examples of that happening. One of my favorite ones from relatively recent history is the move from CVS/SVN in version control to git. When a tool is clearly superior to another it eventually gets adopted. Because of the need to replace, adoption is slow at first (much slower than in a greenfield scenario), but over time it accelerates and eventually a tipping point is reached.

So while the position of Chrome or Gmail may seem unassailable today, there is a distinct possibility that a decade from now we will all be using tools that are in the toy stage right now.