In 2020, I read more books than I have in one year since I was in college. Perhaps the one silver lining of Covid-19 lockdowns. For posterity’s sake, I decided to make a list of my books from the year along with some ratings and notes. Read more...

Servers at Lawrence Berkeley National Lab, home of Cori (the 6th fastest supercomputer in the world)

The last five years has seen a huge proliferation of internet-connected devices, many of which are produced by tiny but fast-growing startups. In general, startups are at a disadvantage compared to incumbent competitors: they do not have as much cash to invest in development, they have a harder time attracting talent, they are not as well known, and they do not have the same sales relationships that established companies do.

But they do have one advantage: being fast growing can help you juice your metrics. In this post I’ll show that if an IOT startup is growing fast enough, its products will appear to have much better availability than they actually do.

Consider a company selling a server that is availabile 99% of the time. In other words, in a given year of 365 days it is only down for \(0.01 \times 365 = 3.65\) days. (Note that this is actually terrible performance in the enterprise computing world—imagine your whole company having to go without access to its Outlook server for nearly 4 days a year!). Read more...

Hey dear readers, a couple of updates. First, I finished up my data science gig at Pure over the summer and started this fall as an MBA student at Stanford. We’re on the quarter system here, and the first quarter (of six) is actually almost done, which means that I can tell you more than you ever wanted to know about revenue recognition 🤓 (but little about anything else).

Second, I decided that it was high time to get started with a project I’ve been meaining to put together for at least the last six months: my ice cream map. The current version has nine of my favorite scoop shops from around the country. Frankly, I was hoping for 10 shops since 10 is a nice, round number. Alas, I could not think of another one that was quite on the same level as these nine, and I didn’t want to force it. Read more...

Growing up, The Simpsons was my favorite television show. Friends was a close second and the radically age-inappropriate Maury reruns my sister and I watched when we got home from school comprised a distant third. There is a delightful compact book out there called The Simpsons: A Complete Guide to Our Favorite Family, which provides more detail than anyone would want on every episode from seasons 1-8. (NB: Ask any Simpsons true believer and s/he will confirm that these episodes were by far the finest of the series.) I somehow persuaded my parents to buy this book for me when I was 10 and subsequently read it from cover to cover, in spite of the fact that all of the information contained in it could be gleaned from owning every DVD and watching every episode multiple times—which I of course did, too.

Fast forward 15 years: I discovered this spring that a kind soul uploaded a massive data dump of 600 Simpsons episodes (episode names, characters, and every damn line from all 600 scripts) to Kaggle. Needless to say, I got pretty excited. Read more...

The electoral college is a disaster for a democracy.

— Donald J. Trump (@realDonaldTrump) November 7, 2012

I’ve recently been doing some work related to political affiliation and voter prediction for a friend who is starting a political advocacy organization. As a result, one thing that I have been thinking about a lot lately is predicting party affiliation for voter targeting. For example, can we use Twitter to figure out if folks are Democrats and target them for get-out-the-vote messaging accordingly? Read more...

I recently ate out with an omnivore friend of mine who asked me before we ordered if I was okay with him having steak for dinner. Read more...

A few weeks ago, I was doing my ML homework and started to wonder how significance testing is applied to ordinary least squares (OLS) regression. I started to do a little bit of poking around and realized that there wasn’t really a good resource that explained, from first principles, how we get from the classic OLS problem to test statistics, p-values, etc. Read more...

My old friend

I live in Berkeley, California, which is known for at least three things:

• Bike theft
• Vegetarian food
• Liberalism

I’ll tackle the first of these topics today; the other two deserve (several) longer posts. Read more...

Up and running! Check back for more later…