Occasioned by the publication of What If?: Serious Scientific Answers to Absurd Hypothetical Questions (see review), an outstanding profile of geek hero Randall Munroe in Rolling Stone.
How much code do you write these days?
Surprisingly, more and more. I do a lot of code just to try answer questions for myself, and sometimes the result turns into a comic. One example that did not turn into a comic — not yet, anyway — I downloaded the whole Google Books Ngrams corpus, and made some tools to visualize whether there were patterns in what years were mentioned in what years. In the 1930s, did people talk about 1776 more than in the 1980s? And you have to normalize for a bunch of things: everyone talks about the current year the most, and the number 2000 gets mentioned a lot, but how much of that was people talking about the year 2000?
I wasn't able to extract a good enough signal from that noise, but I found patterns of mentions of future years in unexpected places. In the 1930s, there was a string of mentions of the 1980s. That seemed weird to me — why are people in the Thirties talking about 50 years in the future? And it was specifically 50 years — 1932 would mention the early 1980s a lot — but the pattern didn't continue in the Forties. I started seeing other ghosts like this that were kind of inexplicable, and then I realized I was seeing OCR text recognition errors, where the 3 was misread as an 8. This was a really effective method to spot the flaws in the text recognition engine they were using. Which was cool, but wasn't I was looking for. I wrote quite a bit of code in the process, making a useful graphic that I could explore. That was a hundred lines of Python, at least, and a couple of afternoons. I do that kind of thing a lot, and occasionally it turns into a comic.
Dropping Science: 'XKCD' Cartoonist Randall Munroe on His New Book
The Stormtrooper Decanter is on back-order, but you can pre-order one from the next batch for £22 — it’s based on Andrew Ainsworth’s original movie helmet moulds from 1976, and will provide endless opportunities to point to lowball glasses and say things like “aren’t you a little short for a Stormtrooper drink?” (via Bonnie Burton)
Yahoo has released a machine-learning model called open_nsfw that is designed to distinguish not-safe-for-work images from worksafe ones. By tweaking the model and combining it with places-CNN, MIT’s scene-recognition model, Gabriel Goh created a bunch of machine-generated scenes that score high for both models — things that aren’t porn, but look porny.
I dote on fidget gadgets — soothing gizmos intended to give your hands something to keep busy with, like modern worry-beads — and while you can’t buy Chris Bathgate’s amazing machined sliders, and the Fidget Cube Kickstarter just closed, there’s still Thinkgeek’s new Jumbo Noah Fidget Toy, which looks like a lot of fun and […]
From self-driving cars to stock market predicting software to the recommendations you get on Amazon and Netflix, machine learning is at the core of modern technology. You could find yourself building technology that is literally changing the world with the skills you’ll learn in The Complete Machine Learning Bundle. This bundle of 10 courses includes 406 lessons that will teach […]
This Python Mega Course will help you learn to code by teaching you to build 10 real-world apps that each highlight a unique use of Python.Job prospects for coders are still growing steadily—and with Python being one of the most popular coding languages out there today, it’s important for job seekers to demonstrate a widespread understanding of the […]
The Atmos R2 may be bigger than the brand’s previously-released vapes, but we argue that in this case it’s definitely a good thing. A bigger heating chamber means more room for packing it full. And the bigger battery means longer, more fulfilling vape sessions. In fact, you can use the Atmos R2 for up to about 25 […]