September 06, 2015

When to Sell

I guess a good time to sell is when I start writing blog posts about the stock market, down about 8% since I posted.

I do think that the big problems haven't in China haven't hit yet, though: a pile of (difficult to discern) bad debt related to government influence over the banking system's investment in state-owned enterprises.

There is an overhang of bad investments in China which is on a scale which perhaps the world has never seen before. Entire ghost cities in plain view. When will the bad debt behind these projects unwind?

Posted by David at 01:00 PM | Comments (0)

September 17, 2015

Starting at MIT

A couple decades ago I never finished a CS PhD at Cornell (really, I barely got off the starting block - I left with a masters soon after passing quals). So I've decided to go back to school. This September, I'm starting work on a CS PhD at MIT. It's been almost 20 years since I've taken any classes, and computer science has gone through several sea changes in the intervening years.

I am taking two classes at MIT this term: 6.858 and 6.869. That is, a systems security class, and a vision deep-learning class.

I've devoted the last few years of my life to the idea that it's important to teach people how to program and that the key is to learn how to program intentionally, to understand how to make things with software.

But the takeaway from security and deep learning so far seem to be (1) even if you do your best to make things intentionally, they can be defeated (we're learning a whole spectrum of security exploits); and (2) there are ways to build systems without detailed intention that outperform intentionally designed systems (deep learning is winning).

In this new world, how to people put together software on purpose? How are we supposed to improve, collaborate on, and evaluate what we're doing?

Anyway, it's very fun to go back to school. Lots of interesting stuff going on to learn.

Here is my academic homepage at MIT.

Posted by David at 03:50 PM | Comments (0)