20220427
2022 04 27
“Went” to a really cool series of talks that were at the Hot Topics in Computing Seminar at MIT (description of event here). James Morris was the main speaker and I cannot recall if he was mentioned in The Dream Machine or not (he was! I just paged through my copy and it took less than a minute to find a reference to him). He gave a really interesting talk on what he thought made breakthrough research in computing and I took a lot of notes on it. One thing that stood out to me was how he thought it was important to pay attention to exponentials and ride the wave with them. I have seen this before with Moore’s Law and Metcalfe’s Law, but I don’t think I have seen someone talk about how both of them in tandem enabled the recent AI breakthroughs: Moore’s Law gave the compute power necessary and Metcalfe’s Law gave the data available on the internet needed to train these models.
He also talked about Second System Syndrome, which is when some group makes a successful system they get a lot of money to make a version 2 that is better in every way, and these second systems almost always fail due to overreach. It’s a classic tale of flying to close to the sun it seems and is important to learn from.
Butler Lampson (another Dream Machine name) also talked and the two principles he built his mini talk around were 1) use the product you are building to put yourself in the feedback loop of improving the product and understanding how changes you make affect the usability of it, and 2) some things just take time to come to fruition. A good idea might exist but it may not have the right environment for it. This was true of AI in the 80s.
Dave Gifford then talked very briefly and the one thing I want to highlight that he said is that it is a very good skill to develop to be able to see “false failures”. To know when something failed because the environment was bad, as Butler said, and did not fail because the idea was bad. Conversely, you want to be able to know if the failure was because the idea was just plain bad too! It can’t always be that the environment was bad.
Daniela Rus was the last speaker and she gave examples of how autonomous vehicles have been and continue to be an idea that is good, but the environment is still not there for it. She said that robo taxis are a long way off and that you should not trust your Tesla autopilot interestingly (not sure what to make of that since I don’t know if there is any bad blood here). She also said that new big ideas just take time to come to fruition and take years of work to realize.
All in all it was a really great experience and I’m still in awe that I’m alive to overlap with these pioneers of computing and not just be able to listen to recordings of them, but to listen live to them (granted I was virtual, but I could have biked 20 minutes to MIT!). It’s been really special to have these experiences for me. Partly because I look up to these people as great inspirations for what can be done, but also for how short of time these things were done in. It’s inspiring because it makes you want to imagine what sort of things we will be able to do in our lifetimes building off of their foundations given that they were able to do so much in so little time. Doug Englebart’s “Mother of all demos” is always referenced at things like this and it makes me wonder what a mother of all demos would look like today to showcase what will be commonplace 60 years from now.
Daily Listening
Still just a lot of Magdalena Bay. I don’t like this album nearly as much, but it still has some hints of what was to come for them.
Daily Reading
Finally back to a little reading - read some of The Rings of Saturn before passing out as usual.