notes.husk.org

Pixel Is Data: What if pixels weren’t necessarily...

Tue, 14 May 2013 22:57:00 +0100

What if pixels weren’t necessarily supposed to look like little squares and sit in the so-called “right order”? What if what we call “real” or “true” images were not the only way the World around us can be represented? What if photographic data was just… data? What if it could be reinterpreted?

Free on the App Store. Images from Fast Co Design, via George Oates.

Amazon and CIA ink cloud deal

Wed, 20 Mar 2013 17:59:00 +0000

Amazon and CIA ink cloud deal:

From the newspaper Federal Computer Week:

In a move sure to send ripples through the federal IT community, FCW has learned that the CIA has agreed to a cloud computing contract with electronic commerce giant Amazon, worth up to $600 million over 10 years.

Amazon Web Services will help the intelligence agency build a private cloud infrastructure that helps the agency keep up with emerging technologies like big data in a cost-effective manner not possible under the CIA’s previous cloud efforts, sources told FCW.

via infoneer-pulse, notational

toffeemilkshake: Every meteorite fall on earth mapped It’s...

Tue, 19 Feb 2013 20:23:26 +0000

toffeemilkshake:

Every meteorite fall on earth mapped

It’s amazing how metorites seem attracted to centers of human population and never ever fall in the sea.

Of course, the name of the graphic should be “every recorded meteorite fall on earth, mapped”.

See also: Astronomers lose access to military data, Nature, June 2009:

The change is a blow to the astronomers and planetary scientists who used the information to track space rocks, especially those that burn up over the oceans or in other remote locations. “These systems are extremely useful,” says Peter Brown, an astronomer at the University of Western Ontario in London, Canada. “I think the scientific community benefited enormously.”

The USAF later said the data would be made available again, but I don’t know if that happened. If it did, it may well be a better source than the data the Guardian worked from.

London Snow via the medium of Flickr

Mon, 21 Jan 2013 00:01:00 +0000

I wondered this morning how common snow’s been recently in London. After all, this year’s looking like having a week or so of snow, and I remember my last winter there, 2010, being fairly white too.

I was also reminded of this by Boris Johnson’s recent (generally awful) Telegraph column, where he says

By my calculations, this is now the fifth year in a row that we have had an unusual amount of snow; and by unusual I mean snow of a kind that I don’t remember from my childhood: snow that comes one day, and then sticks around for a couple of days, followed by more.

OK then. I suppose I could double check by looking at the Met Office’s UK climate summaries, but that would require some reading comprehension, and it’s a Sunday. Instead, I thought I’d do a tiny bit of data mining. (Actually this hardly qualifies, but what the hell, big data’s sexy, right?)

Flickr have an API, and one of the core methods is flickr.photos.search, and one of the parameters is the date taken. So it’s pretty trivial to write a small Python script that will do the search, return the total count for a search for, say, ‘snow london -ontario’, compare it with a baseline of ‘london -ontario’, and get this:

2001	34	20505	0.165813
2002	206	46747	0.440670
2003	419	90416	0.463414
2004	763	187478	0.406981
2005	1879	515875	0.364236
2006	2551	1130056	0.225741
2007	15227	1838767	0.828109
2008	12192	2027861	0.601225
2009	64871	2326955	2.787806
2010	34149	2305502	1.481196
2011	7429	2322795	0.319830
2012	14241	2449517	0.581380
2013	4872	63543	7.667249

Only three years reach over 1% of ‘snow’ photos, by this (admittedly handwaving) method: 2013, 2010, and 2009 (which was actually snowier, by this measure). By contrast, 2011 and 2012 look far less snowy.

(Of course, 2013 is pretty biased, because we haven’t had the non-snowy months that a full year has.)

Now I’ve produced this, I should actually go and do the hard work of comparing it to the aforementioned summaries to see if it’s actually worthwhile or not.

Edit: hugovk suggested looking for winters rather than years, so I changed the start/end of the timekeeping period to be in September of the year shown. Now the results look like:

2001	119	35647	0.333829
2002	443	73337	0.604061
2003	594	153578	0.386774
2004	1587	377460	0.420442
2005	2027	898777	0.225529
2006	15329	1671273	0.917205
2007	10903	1989473	0.548035
2008	60843	2250467	2.703572
2009	20579	2295751	0.896395
2010	25089	2316599	1.083010
2011	14916	2502527	0.596038
2012	6921	764047	0.905834

This looks better for the year starting in September 2012, and also makes 2006/2007 and 2009/2010 come up towards that 1% limit. Better.

frankie-roberto: Every time I see a ‘trig point’, I can’t...

Tue, 15 Jan 2013 04:44:00 +0000

frankie-roberto:

Every time I see a ‘trig point’, I can’t resist the urge to photograph it.

Something about trig points makes me want to try and visit them all. This would probably be a ridiculously tall challenge. I bet someone’s done it though.

Geographical Magazine, March 2009:

Rob Woodall, 48, is attempting to become the first person to visit all 6,100 surviving Ordnance Survey triangulation pillars in Britain. For this exploratory challenge he now has fewer than 200 to visit.

Gnip, Twitter, and Archives

Mon, 24 Sep 2012 16:27:37 +0100

notational:

Social Data Startup Gnip Goes Deeper Into Twitter’s Past, Offers Full Archive Of Public Tweets
courtenaybird:

Gnip is announcing a new product today that provides access to the full database of public tweets from the beginning of time — or rather, the beginning of Twitter.

…The new product, called Historical PowerTrack for Twitter, has been in testing with customers including Esri, Brandwatch, PayPal, Brandwatch, Waggener Edstrom, Network Insights, Union Metrics. Moody says this data opens up a number of new use cases. For one thing, financial firms are developing trading algorithms that incorporate Twitter data, and they can now test those algorithms on data from the past — in other words, if they think they can use social network activity to predict of stock market activity, they now have a giant database for seeing whether that’s true. Moody says there are also academic researchers looking at the impact of Twitter activity on the Arab Spring.

“We fundamentally believe that social data is going to be in every application. We’re only at 1 percent of the journey.”

(via TechCrunch)
I can’t download my historic posts beyond the most recent 3,200, but Gnip can (presumably with Twitter’s full blessing) sell access to people wanting to use sentiment analysis for stock trading.

CityDashboard: London (via), by the CASA research lab at...

Sat, 28 Apr 2012 14:18:07 +0100

CityDashboard: London (via), by the CASA research lab at University College London.

There’s a fuller list of contributors and sources on the about page, along with this disclaimer:

CityDashboard is an early prototype and should be considered to be “alpha quality” - expect data feeds to break regularly. Please do not rely on information display in CityDashboard, as it may be erroneous. For example, if the CASA Geiger counter is showing a high reading, please do not panic! Somebody in the office might just have placed some Brazil nuts or another calibration source in front of the detector.

Four of the many, many different map overlays at the London...

Tue, 24 Apr 2012 05:48:43 +0100

Census classifications

Violence against the person

Travelcard zones

Welsh population, output area level

Four of the many, many different map overlays at the London Profiler site. Sadly the data is a little out of date (it looks like it hasn’t been updated since around 2008; I first saw it in 2009), but it’s still fun to play with for a while.

(The images are captioned, but if you’re really curious about what’s displayed, I’d urge you to check out the maps in full.)

“Map of City of New York, Showing the Distribution of the...

Tue, 27 Mar 2012 14:25:05 +0100

“Map of City of New York, Showing the Distribution of the Principal Nationalities by Sanitary Districts” (1895). Each pattern reflects a different ethnicity. (from, via)

How Many Photos Do Americans Take a Year? on Hyperallergic,...

Fri, 23 Mar 2012 21:33:00 +0000

How Many Photos Do Americans Take a Year? on Hyperallergic, referencing the April edition of National Geographic.

My response to that is, “so few”? I can take 255 photos on a (very) good day out, but then, I suppose not everyone has a camera, nor do they photograph everything.

Also filed under “so few?”:

Last year, 37% of the images in the US were captured using camera phones, but by 2015, National Geographic writes, that number is expected to be 50%.