According The Intercept (and implied by The Washington Post), NSA’s MYSTIC was/is storing only metadata of mobile-phone calls in Kenya, Mexico, and the Philippines and recording metadata, and the actual voice call, in the Bahamas and a fifth, unnamed country.
Later that day, WikiLeaks announced they will name the fifth country. The Intercept’s article, however, purports to show an image of the memo naming the country but with the country’s name covered with a black box. If that image is accurate, then it might be possible to determine which country is redacted in the image.
The image of the memo is in JPEG format, which is usually used for photographs not text. Because JPEG tries to smooth all lines and is lossy format (data is lost when it is compressed and saved), the letters in the image were too mangled for the automatic font finders at MyFonts or WhatFontIs to locate a good candidate for the font used in the original memo.
Identifont asks a series of questions to narrow the field of possible fonts. Many of the questions were unanswerable, such as descriptions of the upper-case Q letter, because the letter, glyph, or number was not in the memo image. After 37 questions, however, Identifont suggested URW Latino.
The font has enough distinct features that it seems likely that the non-bold text in the above image uses URW Latino. The above image has at least two fonts in it: notice the lower-case g is different in the bold, italics text than the normal text.
If URW Latino is the correct font, we know almost exactly what the em/ex length of the redacted word(s). With the font test tool at URW, Pakistan is too short, and Afghanistan seems to be too long by about 90% of the final “n” length. Too long: Bangladesh, Madagascar, Puerto Rico, Switzerland, and Venezuela. Too short: Columbia, Hong Kong, Ivory Coast, and Kazakhstan. Possibly: Guatemala, Macedonia, and Mauritania.
If I try other countries, or if I do more than an eyeball measurement on country names, variations, and abbreviations, I will update the post.