Profile cover photo
Profile photo
Thomas Colthurst
Thomas's posts

Post has attachment
A critical review of Gary Taubes's "The Case Against Sugar". Recommended. Hat tip to

+Karen MacArthur

Post has attachment
In Barcelona. The green birds are parakeets, which describes as "Most notable among non-native species. It bred in the city for the first time in 1974 from a group of escapees. They now reign over the whole city. Although they add a streak of colour and fun to the city’s dull litany of pigeons, sparrows and seagulls, they can cause serious problems as they expand out into the country."

Two of the photos are of La Sagrada Familia, which is the most beautiful place I've ever been.
4 Photos - View album

Post has attachment
TAP Portugal lost my bag, but luckily I was able to restock my witticisms here.

Post has attachment
You know that "Oh yeah" early-electronica song from Ferris Bueller's Day Off? This is the original video for that song. Recommended, especially if your day has otherwise lacked WTFs.

Post has attachment
I'm really hoping the tattoo artist didn't charge him, because then it would be a free Nelson mandala.

HT to

Post has attachment
Dear World,

Could you stop sending me links of this article with the "But the kinds of people who work at Google are often the ones who became software engineers because they wanted to avoid talking about feelings in the first place." sentence highlighted?


Your humble and eternal servant,

Post has shared content

Post has attachment
This is what I've been working on for the past year.  Very excited for it to have finally launched!

Post has shared content
Like The Beatles in 1966 or Western painting in the 1870s, neural networks have moved beyond mastery and are now entering their psychedelic phase.

And I never thought I would have to say this in a post about machine learning, but:  WARNING!  SOME OF THESE PICTURES FEATURE EYEBALLS EVERYWHERE AND MAY CAUSE NIGHTMARES.
Artificial Neural Networks have spurred remarkable recent progress in image classification and speech recognition. But even though these are very useful tools based on well-known mathematical methods, we actually understand surprisingly little of why certain models work and others don’t.

Over on the Google Research blog, we take a look at some simple techniques for peeking inside these networks, yielding a qualitative sense of the level of abstraction that particular layers of neural networks have achieved in their understanding of images. This helps us visualize how neural networks are able to carry out difficult classification tasks, improve network architecture, and check what the network has learned during training. 

It also makes us wonder whether neural networks could become a tool for artists—a new way to remix visual concepts—or perhaps even shed a little light on the roots of the creative process in general.

Post has attachment
For every dollar we spend on Medicaid, what is the value of the benefits that the recipient gets?

A new paper uses the Oregon Health Insurance Experiment (see and for my previous posts on the experiment) to answer that question.  It's called, naturally enough, "The Value of Medicaid:  Interpreting Results from the Oregon Health Insurance Experiment".

The answer in the paper's abstract is "$0.2 to $0.4", but they also do a sensitivity analysis using a variety of "reasonable" assumptions (their scare quotes, not mine) that gives a range from $0.15 to $0.85.  The authors (Amy Finkelstein, Nathaniel Hendren, and Erzo Luttmer) note that these estimates are in line with previous measurements of how much Medicaid recipients would be willing to pay for their benefits, which give values from $0.3 to $0.5.

The astute reader will notice that all of these amounts are less than $1.

HT to Marginal Revolution.
Wait while more posts are being loaded