I remember once flew to meetings in other countries and working with a group of people to annotate the proposed criteria. The convener projected the word documents onto the screen, and people called for the proposed changes. It was discussed in the room before being adopted or adapted, added or subtracted. I'm not a child.
I don't remember exactly when this was, but I know it was after Google Docs was introduced in 2005. Because I remember that this international standard organization was completely confused in the last century that it was still trapped somewhere.
You may not have experienced anything about this extreme, but many remember the day when you send a word file as an attachment and collate and compare multiple different versions. This behavior also continued after 2005 (obviously this still applies in some contexts, such as some of the US government). If you're not as old as you've experienced it, consider yourself lucky.
This is, in many respects, the point of Arvind Narayanan and Sayash Kapoor's essay “AI as normal Technology.” There is a long gap between the invention of technology and the true understanding of how to apply it. One standard example was at the end of the second industrial revolution. When it was first electrified, the factory replicated the design of the factory, loaded with coal and steam. There, a huge central boiler and steam engine distributed mechanical power across a variety of machines through the complex arrangement of gears and pulleys. The steam engine was replaced by a larger electric motor, but the factory layout remained unchanged.
Over time, it was reconfigured to utilize small electric motors that could be distributed throughout the factory and incorporated into individual specialized machines. As I discussed with Arvind Narayanan last week, there are four stages of any technological revolution. The invention of new technology. Spreading knowledge about it. Development of products based on this. Adaptation by consumers, businesses and society as a whole. This will take time. I love James Bessen's love for framing this process as “learning by doing.” It takes time and shared learning to understand the best ways to apply new technologies and search for their possibilities. People try new things, show them to others, and build on them with a wonderful leap of imagination.
So it's not surprising that the 2005 file was still emailed. One day I came up with a way to realize the true potential of the Internet and create an environment where all the mechanisms of version control exist but hidden from sight can share files in real time.
Next Tuesday episode Living with Tim O'ReillyI talk to that small group (Sam Schillas, Steve Newman, Claudia Carpenter). Writely was acquired by Google in March 2006 and became the basis for Google Docs.
In the same year, Google reinvented online maps, spreadsheets, and more. It was in the year that, since the early 1990s, basic lessons on the internet quickly began to sink.
It is very important to remember this moment. Because we are at a similar point today. We know what to do with AI, but rather than truly searching for the possibilities of deployed features, we are building our factory equivalents with a huge, concentrated engine. Ethan Morrick recently wrote a great essay about “bitter lessons with the trash can” about the opportunity of this moment (and failure mode). Do we really start to figure out what is possible with AI, or do we try to fit it into old business processes? We must wrestling with the angels of possibilities and remake familiar things into what we can now vaguely imagine.
I am truly looking forward to talking to Sam, Steve, Claudia and everyone in attendance, not just to reflect on their achievements 20 years ago, but also what we can teach us about the present moment. We would appreciate your participation.
AI tools are moving rapidly beyond chat UX to sophisticated agent interaction. Upcoming AI Codecon events, Coding the Agent Worldhighlights how developers are already using agents to build innovative and effective AI-powered experiences. I hope to join September 9th to explore the tools, workflows and architectures that will define the next era of programming. You can attend freely. Sign up now and save your seat.
