How Allies Used Math Against German Tanks

If you care to search you'll find that this story has appeared here on HN several times in the past several months. There has been much discussion, again and again. I'm currently on my mobile and can't easily search for you.

EDIT:

OK, I've got to a proper computer and here are some search results:

http://news.ycombinator.com/item?id=670218

http://news.ycombinator.com/item?id=1421698

http://news.ycombinator.com/item?id=1818166

Each of these has substantial discussion, and I'm pretty sure there are more.

If you like stories like this, there are a number of similar stories (e.g. this one, cracking the enigma code, protecting convoys) in a book called The Pleasures of Counting: http://www.amazon.com/Pleasures-Counting-T-W-K%C3%B6rner/dp/... - it's not all world war 2, but is very interesting and does contain a lot of information on the math used during the war.

When I saw the headline, I my first thought was that the math was something along the lines of:

A = number of U.S./allied tanks in each battle

D = number of German/axis tanks in each battle

A >= 2D

or why to not put autoincrement id's into your webapp routes

...because others then can find out how many visitors your site gets? Or perhaps nastier things, like guessing id's?

Previously: http://news.ycombinator.com/item?id=670218, http://news.ycombinator.com/item?id=1818166, joshu: http://joshua.schachter.org/2007/01/autoincrement.html (http://news.ycombinator.com/item?id=52429 )

Foursquare growth in 2009 inferred from autoincrement IDs: http://www.flickr.com/photos/wallofhair/4232092914/

I have a bunch of similar data on craigslist I should write up someday.

One from 2006:

http://www.guardian.co.uk/world/2006/jul/20/secondworldwar.t...

"The statisticians believed that the Germans, being Germans, had logically numbered their tanks in the order in which they were produced."

For some services, your growth may not be proprietary info — you may want to show incrementing IDs off, or make all your artifacts trivially discoverable.

But, if you have an internal autoincrement ID that you want to obfuscate, what's the best simple practice, minimizing token expansion?

Just use UUIDs and be done with it. I'll give you a $1 if you ever see a collision ;-)

That's a reasonable baseline suggestion, but I'd prefer any technique to generate more compact public IDs.

http://www.quora.com/How-do-you-symmetrically-encrypt-32-bit...