Photo by Stefano

Seven command-line tools for data science – 90% of data science is loading the damn stuff, and this is a great set of basic utilities for a lot of the formats you’ll have to deal with.

Classifying digits with deep-belief networks – A very readable guide to the new new thing in machine learning!

Our logo looks like underpants – British people are weird.

Busting the King’s Gambit, this time for sure – I don’t know chess, but the state of the art of computerized analysis is amazing.

Sloane’s Gap – A numerical investigation into strange properties of a large collection of number series. I learned that 11630 is the first uninteresting number, and there are 350 interesting sequences that contain 1729, so it’s even more exciting than Ramanujan thought!

Picture by Nicholas Raymond

I’ve published a new version of the Data Science Toolkit, which includes David Blackman’s awesome TwoFishes city-level geocoder. Largely based on data from the Geonames project, the biggest improvement is that the Google-style geocoder now handles millions of places around the world in hundreds of languages:

http://www.datasciencetoolkit.org/maps/api/geocode/json?sensor=false&address=القاهرة

{
  "status": "OK",
  "results": [
    {
      "types": [
        "locality",
        "political"
      ],
      "address_components": [
        {
          "types": [
            "locality",
            "political"
          ],
          "short_name": "Cairo",
          "long_name": "Cairo, EG"
        },
        {
          "types": [
            "country",
            "political"
          ],
          "short_name": "EG",
          "long_name": "Egypt"
        }
      ],
      "geometry": {
        "viewport": {
          "southwest": {
            "lng": 31.1625480652,
            "lat": 29.9635601044
          },
          "northeast": {
            "lng": 31.3563537598,
            "lat": 30.1480960846
          }
        },
        "location": {
          "lng": 31.24967,
          "lat": 30.06263
        },
        "location_type": "APPROXIMATE"
      }
    }
  ]
}

You can also access the TwoFishes API directly, which offers a lot of very powerful features like breaking down search queries into their where and what parts, so you can do something useful with “Pizza New York”.

For the first time I’ve made AMIs available in all the EC2 regions worldwide, and you can download or torrent the Vagrant version. Have fun, and let me know about any issues or improvements you’d like to see!

Photo by Racineur

Software development without estimates, specs, or other lies – The secret to being a good coder is understanding the business problem you’re being paid to solve. I know, you just want to code, but that skill’s getting democratized to death. Your real value manifests when you hunt down a ton of messy and contradictory needs and figure out a solution that works for the most important ones.

The mystery of San Francisco English – Did you know San Franciscans used to sound like Brooklynites?

On Chomsky and the two cultures of statistical learning – Peter Norvig dismantles Chomsky’s dismissal of statistical models. Bring popcorn.

Five lies your world map told you – Country borders aren’t nearly as well-defined as you might think. Some great examples in here, including the Bir Tawil Trapezoid that neither Sudan nor Egypt want to claim.

If it doesn’t work on mobile, it doesn’t work – Must-read data on understanding mobile. Written by Brian Boyer from the trenches of web development, you’ll learn why almost no mobile use is by people moving, that they prefer reading on phones to desktops or even tablets, and that American hourly usage patterns are very similar to British, but without a tea-time spike at 4pm!

Photo by Yasa

After my post last week about OpenHeatMap’s removal from Github, I’m very pleased to say that it’s back up! The Github support team got in touch quickly after my post. We never did figure out what happened with my emails, but they worked hard with me over the weekend to get all the file removal issues straightened out. As I mentioned in the original article, I’ve been fond of Github for a long time, so I’m very pleased to that we could get this sorted. Thanks to John Greet especially for collaborating with me on the technical side of the removals, on a Saturday afternoon!

I’ve also had a lot of questions about why I didn’t just move to an alternative provider like BitBucket? OpenHeatMap is a niche project that’s only useful to a small, scattered group of people. I open-sourced it so as many of those folks as possible could find it, and in my experience Github’s overall popularity and social hooks make it the best place for projects like that to be discovered. I do think there’s a future for self-hosted alternatives, the way WordPress works with commercial and open off-shoots, but projects like Gitlab still feel very clunky compared to Github right now.

Photo by Etienne Girardet

TwoFishes – A kick-ass geocoder for everything above the street level, by David Blackman based on geonames and other public data. I’m hoping to roll this into the next distribution of the DSTK, it does a brilliant job handling a lot of tricky problems like native-language unicode.

NIST Randomness Beacon – Provides a new set of random numbers every sixty seconds, along with a checksum linking them to the previous randoms, and an archive of all the random values from previous time intervals. I have no idea what to do with this, but it feels like such an interesting primitive for applications that need verifiable timestamps.

Understanding DMA Malware – After hiding code on hard drive controllers, and then in CPU’s microcode, here’s an example of writing a keylogger that runs entirely through the direct-memory access controllers that most systems support, with demos on both Linux and Windows. As our devices become saturated with computation, there are so many places for malicious code to hide.

What are these giant concrete arrows across the American landscape? – We used to need a chain of massive earthworks across the continent to help planes navigate!

Mapping and its discontents – I’m excited to see Berkeley focusing on the power of maps as stories rather than treating them as a technical subject. I hope I can make this symposium, I’d love to hear from folks like Rebecca Solnit, author of the wonderful Infinite City atlas of SF.

Photo by Sinead Fenton

OpenHeatMap is an abject failure commercially, but I’ve kept it running because around 40,000 people a year use it to create simple visualizations. It’s free and aimed at non-technical folks, so a lot of them are from non-profits, schools, local activist groups, and other causes that I enjoy helping.

I spend a few hours a week answering emails, and occasionally have to do some server maintenance, but it’s generally a fairly light labor-of-love. The source code has always been up on Github, but a few months ago it was taken down after a copyright complaint. Unfortunately Github don’t seem to have any process for me to contest this decision, and after a few inconclusive exchanges with their support team supplying further information they’ve stopped replying to my messages.

It’s not really hurting me personally, I’m still able to keep up my maintenance on the project from a local copy, but I’ve had to field puzzled emails from people who want to fork or learn from the project, including one team with an intriguing idea for a suite of visualization tools aimed at non-profits. I don’t like disappointing these folks, so I’m putting this together as an explanation of why I’ve not been able to help them.

The actual complaint isn’t completely nuts, but what has left me sad is how Github handled it. About five years ago somebody emailed me a bug report, attaching a few spreadsheets of addresses and names, without any description of what they were. This is a pretty typical use case for the website, I’ll often see what look like short fragments of a phone book as test files. I fixed the bug and added the file to my unit tests. A few months ago, years after I’d originally created the unit test, I received an email from a CTO at a consulting firm, angry that what appeared to be a list of his staff was available if you dug around enough on Github, though it was unlabeled. It appeared one of his employees had sent it to me as a test case. I felt a bit embarrassed – looking back I’d be more careful about what I used as my unit test inputs, at least scrubbing them more vigorously so that only address data remains (though a lot of the tests are about identifying what is address data in a soup of other columns). I went ahead and removed the offending data from my repository and its history, and checked the result in. I also started the process of removing the unit test directory entirely from the public project, though that was a longer task. Unfortunately the complainant found another copy of the file he hadn’t spotted before, and rather than contacting me, got in touch with Github and persuaded them to disable the project.

I was pretty disappointed, but assumed I could get the project back online with the unit tests removed entirely. Unfortunately they don’t seem to have any kind of process for resolving problems like these, and the only point of contact I’ve had is through their main support@github.com email. I tried reaching out on Twitter too, but without any luck. Right now the project’s stuck in limbo, apparently permanently banned, and I’m not sure how to get it online again. I’m a long-time fan of Github, and there’s no other provider that offers such a good environment for sharing source code, so I’m just sad that they don’t seem set up to handle this sort of problem, and I hope this doesn’t affect other projects going forward.

[Update – I’ve had an email back from Github, it sounds like my previous mails may have gone astray, and they’re on the case.]

[Update 2 – OpenHeatMap is now back online, thanks to everyone for their support!]

Photo by Dustin Jamison

A friend just asked “How popular are the links on your Five Short Links posts compared with your ‘regular’ posts?“. This seemed like a good chance to share some data, so here’s a rundown of who reads what, and what drives me to write the posts.

I get 25,000 unique visitors on a typical month, and around 35,000 views. I also have around 4,000 RSS subscribers according to Feedburner, but I don’t know how many of those are actively reading my blog. My biggest traffic sources are search engines, then Twitter, Hacker News, and Reddit. I’ve found I write three kinds of posts, and their traffic patterns are very different.

Show and Tell

When I’ve got a topic I want to tell the world about, I’ll put together a post with some examples and a bit of background on why I think it’s important. A lot of these end up with just a few hundred views, maybe a few thousand if they end up shared on Twitter, and occasionally I’ll end up with tens of thousands if I really hit a nerve. My recent post on Google’s geo APIs has racked up 42,349 views over the last few weeks, largely thanks to a stint on the front page of Hacker News. My post on distrusting data scientists never made it onto a big aggregator like that, but was widely shared on Twitter and through specialized blogs, and has had 13,792 views since it was published. The post I did on name analysis is more typical, with 2,871 views since it was published.

These articles usually take quite a lot of time to research and put together, which means I can rarely do more than one or two a month. I write them because I can’t help myself! I’m passionate about my work, and I love having a platform I can use to grab people by the lapels and rant at them.

How-to’s

It takes less time to write up my notes on a technical problem I’ve had to figure out. I think of these as trails of breadcrumbs through the forest, and my hope is that anyone who  else who hits the issue can Google it and find something helpful. I rely on other people’s write-ups as a starting point for almost any bug I hit, so my goal is to pay back some of that help, and keep the ecosystem of user-written documentation alive. A typical example would be my post on debugging Javascript errors on iOS which gets 1,932 views a year. These tend to be evergreen, keeping steady traffic for years, and almost everyone finds them through search engines.

Links

The easiest posts, and so the most frequent, are my short link digests. I have a large list of blogs I follow through Feedly, and I often run across interesting articles while I’m searching on technical topics, and from the folks I follow on Twitter. I also find the ‘newest’ page of Hacker News full of neglected gems. A lot of my favorite links never make it to the front page, which is usually heavy on controversy and unkind to interesting-but-unsensational projects.

I’ve been collecting links for years,  and copying Nat Torkington’s Radar post format (plus 25% extra) gave me a fun way to share them with the world. The posts don’t get a massive number of clicks, the last three posts got 66, 61, and 91 views total, but people seem to like them. I end up having a lot of conversations with folks I never would have been in touch with, and it feels good to shine a light on projects that deserve more attention. My favorite result is seeing a startup or framework get picked up by a publication with a much bigger audience, since I seem to have a decent number of journalists and other bloggers following my posts.

Photo by Romana Klee

The joy of unrepresentative samples – It’s uncontroversial in the commercial world that biased samples can still produce useful results, as long as you are careful. There are techniques that help you understand your sample, like bootstrapping, and we’re lucky enough to have frequent external validation because we’re almost always measuring so we can make changes, and then we see if they work according to our models. The comments on this post are worth reading because the approach seems to offend some sociologists viscerally. (via Trey Causey and Benjamin Lind)

Humanize – A Javascript library that handles the common language transformations like translating numbers into positional text (eg 1 into ‘first’), turning lists into comma-separated strings with ‘and’ between the last two entries, and other goodies. I wonder if this will be translated into languages other than English?

Should Excel spreadsheets be subject to external peer review? – Making it easy to get more sets of eyes on your data.

Thoughts on Intel’s upcoming software guard extensions – My conclusion after reading this overview is that the complexity of modern processors is mind-boggling, and it’s becoming increasingly impossible to verify the security of any of the hardware or software we use by inspection.

Black Midi – Jamming insane numbers of notes into an ancient music format, and playing them back with the dinkiest software you can find. A thing of beauty.

Photo by Chintermeyer

Black Perl – “BEFOREHAND: close door, each window & exit; wait until time. / open spellbook, study, select it, confess, tell, deny;” – A compilable poem, beautifully weird.

Relationship Timelines – It’s rare that a network visualization illuminates, rather than impresses, but XKCD’s Lord of the Rings et al narrative charts actually added something to my understanding of the movies. Skye Bender-deMoll has pulled together some similar research examples to try to figure out how to create similar graphs automatically, and I’m hoping he succeeds.

A deadly gift from the stars – A meditation on the fragility and unlikeliness of life, built around Iain Banks’ hope that his cancer was caused by cosmic rays rather than something more banal.

Freedreno updates – I spent months working with the then-ATI driver engineers debugging problems with the way we used their Radeon GPUs at Apple, but I was never able to see their source code. That means I’m excited to see an open-source driver for what looks like a very similar chip, the Adreno, used on a lot of mobile devices. The development process is fascinating too, I wish I’d been able to instrument the drivers to understand performance problems in the depth Rob is able to.

GitSpatial – Github are making a big effort with their GeoJSON support, which is an interesting expansion outside of their traditional code focus and into data. GitSpatial is an intriguing layer on top of that support, adding a query API with Github as the backing store.

Photo by Tambako the Jaguar

Where’s my fusion reactor? – An engrossing overview of the state of smaller fusion research projects. For the past half-century, fusion has permanently been twenty years away, so I’d love one of these to come out of the shadows and surprise us all.

Smathermather’s weblog – I don’t often link to entire blogs, but Stephen Mather’s is so full of impressive geo-hacking posts it would be an injustice to link to just one of them. I am particularly fond of his use of POV-Ray for analyzing the available views from particular points in the landscape though. I spent the summer of 1990 furiously rendering 160×120 images using POV trying to create the ultimate mirror-ball on a chess-board. It left me amazed that there were programmers were generous enough to give the software away for free, and itching to write something myself.

Finding important words in a document using TF/IDF – A straightforward explanation of a powerful approach that’s often cloaked in jargon.

Unusually effective debugging – Early in my career I noticed that I spent most of my time debugging, and that the biggest difference between the most productive programmers and the least was how effective they were at it. You end up debugging when there’s a mismatch between the mental model of what you think your code should be doing, and how it’s actually being executed. This article has some excellent advice on ways to find the flaw in your mental model as quickly as possible: “It’s about killing your darlings, looking for evidence to prove your theories false. It’s about ignoring the how and why and describing, as precisely as possible, what the problem is. It’s about imagining a huge multidimensional search space of possibilities and looking for ways to eliminate half or whole dimensions, recursively, until you’ve isolated the fault.”

Akkie, and the 101 things you can do with a CD-ROM drive’s eject function – There’s a zen-like beauty about focusing on the possibilities of misusing a single basic component in creative ways. Feeding hamsters, twitter notifications, ringing bells, all pure hacks in the best way possible.