Wikipedia:Reference desk/Archives/Computing/2014 November 10

Welcome to the Wikipedia Computing Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

November 10

Flikr too big

I'm trying to access Flikr with Palemoon, I see only the top left quarter of what I should see once I select an image. The usual <ctrl>- does not help. All the best: Rich Farmbrough, 00:33, 9 November 2014 (UTC).

Bouvet email addresses?

I subscribe to a listserv that just got meta-spammed — someone sent an advertisement for a spamming firm, listing how many email addresses they had per country and what they'd charge to spam each of them. One entry reads as follows:

Bouvet Island 1602 Email Addresses $50 USD

I can vaguely understand BIOT addresses (perhaps they're for workers and military offices on Diego Garcia), but Bouvet, Heard Island and McDonald Islands, and the Territoire des Terres australes et antarctiques françaises are all uninhabited (and the USSR is defunct), but all of them supposedly have email addresses. Do they really (and if so, how?), or is the spammer branching out into defrauding its customers? Nyttend (talk) 16:04, 10 November 2014 (UTC)

ISO 3166-1 (list of country codes) has "BV" for Bouvet Island, "IO" for British Indian Ocean Territory, "HM" for Heard Island and McDonald Islands, and "TF" for French Southern Territories (redirected to French Southern and Antarctic Lands; in French, "Terres australes et antarctiques françaises" or "Territoire des Terres australes et antarctiques françaises").

—Wavelength (talk) 17:11, 10 November 2014 (UTC)

Well, as you already know, they're a meta-spammer - which means that they're aggregating data, which is by its very nature unstructured data, from untrustworthy sources. One of the biggest challenges of so-called "big data" is separating wheat from chaff: in other words, identifying invalid entries. Most probably, the list of "compromised emails" comes from a giant geo-tagged database, and that data contains many invalid entries. Only the most egregious and obviously wrong entries are easy to spot - so you (as a human) have a sort of immediate ability to filter out junk with high probability. But the aggregation algorithm - which, by its very nature, is run on a spambot - can't do that! Identifying certain categories of junk data-entries in an email list is a great example of a task that would be obvious and trivial for a human, yet difficult to program as a spambot-algorithm.

If I were trying to make money selling spam, I'd intentionally avoid sending the spam (it's too easy to get busted!), and I'd instead opt to monetize selling junk data to spammers, (which entails a lot less legal liability!) Besides, selling giant phony lists of contact information for purportedly-real-humans is a lot more lucrative marketplace than ordinary run-of-the-mill spamming! Nimur (talk) 17:40, 10 November 2014 (UTC)

.bv says the domain is not available for registration so I don't see where they got all those emails from. I'd guess they are untested emails that have bee put in by people in error - or because they don't want to be emailed. Anyway what do you expect from people like that. Dmcq (talk) 17:46, 10 November 2014 (UTC)

Of course there are honeypots of fake addresses created by the "good guys", on their own servers for these people to harvest. If (say) the required response from 1 million spams is 10 victims, then diluting the address pool 1000 fold means these spammers are very unlikely to have a successful business model. For more details mail me: RichyRich@mymailserver.bv. All the best: Rich Farmbrough, 02:40, 11 November 2014 (UTC).

Wpoison is one such tool. Here it is in action. It uses random two-letter TLDs for many of its fake email addresses. -- BenRG (talk) 07:53, 11 November 2014 (UTC)

Are matplotlib, gnuplot, and latex connected at a lower level?

Are they processing the same objects down there?--Senteni (talk) 18:25, 10 November 2014 (UTC)

matplotlib uses GTK+. gnuplot usually uses wxWidgets or Qt for graphical output (or something else, if you've got an unconventional installation). LaTeX is a stand-alone utility that produces a .dvi file suitable for conversion into some other data-type that can be rendered on screen (like a PostScript or PDF file).

Deep in the innards of these programs, all are munging with representations of visualizations that can be lumped into the broad categories of raster graphics and vector graphics. However, these three tools share very little code. In other words, although they all have common types of abstractions about plottable or drawable objects, they do not use the same implementation of these abstractions - internally, their abstractions are not even cross-compatible.

If we wanted to be very pedantic, we could make the case that none of these three programs actually draw anything to screen - that's a job that they all leave to an external graphical toolkit. Most of the time, though, if you download these packages as part of a plotting utility kit, they will come pre-linked and pre-installed with the graphical kit, so that the plots can be written to standard image formats or drawn interactively on your monitor.

Nimur (talk) 18:38, 10 November 2014 (UTC)