What does randomness look like? (2012) | Hacker News Reader

What does randomness look like? (2012)

wired.com

62 pointsxsace20 comments

What isn't mentioned is the fact that British intelligence fed false news article back to the Germans saying that the bombs had fallen short of their target. The Germans then adjusted their flight paths to make the bombs fly further, causing many of them to fly straight over London and land in the countryside the other side.

Actually, the bombs were already falling short of the target. The British instructed captured German spies to report they were hitting their targets. They were quite subtle about how they did it -- they would simply omit news about the bombs that fell short.

My favourite story is about Battle of the Beams (http://en.wikipedia.org/wiki/Battle_of_the_Beams), where German bombers were guided by radio navigation. The British started broadcasting their own radio signals to make it appear to the German planes that the beam was slightly bent. It meant they could get the bombers to drop their bombs in the middle of nowhere.

The last word of the references section of the article ("teaser") links to a more detailed comment about this -- from Hacker News 2 years ago:

https://news.ycombinator.com/item?id=4955201

Both distributions are random, they're just different kinds of random. The one the article refers to as "random" is _uniformly distributed_, while the other is not. Similarly, different _kinds_ of random distributions sound different! Wikipedia has a nice article on various noise distributions with audio samples: http://en.wikipedia.org/wiki/Colors_of_noise

Brown noise sounds lovely, like a waterfall: http://en.wikipedia.org/wiki/Brownian_noise

The British worried about the accuracy of these aerial drones. Were they falling haphazardly over the city, or were they hitting their intended targets?

This is the seed of Thomas Pynchon's 1973 novel Gravity's Rainbow -- a masterful, if difficult, novel about (among other things) paranoia.

I've never made it very far through Gravity's Rainbow (although I have tried multiple times) but I'm pretty sure it's about the ballistic V-2 missiles rather than the cruise-missile like V-1s.

That's true. It's about the V-2s (in particular, one mysterious weapon the Germans appear to be building called the "Schwarzgerät," with serial number 0000).

But that business of trying to figure out whether the bombs are random or not is central to the book. There are a lot of statisticians running around trying to figure out why the pattern of bomb blasts corresponds to the Poisson distribution.

(Of course, there's also a character whose erections appear to predict bomb blasts. Postmodernism, ftw).

I just finished reading GR a couple of months back, I wish I had read this article before... I'll be sure to re-read this article before I read GR the next time...

Gravity's Rainbow is also a tribute to Richard Fariña's "Been Down So Long It Looks Like up To Me". Pynchon and Richard Fariña were friends. The two are very linked conceptually.

BDSLILLUTM is decidedly about paranoia. Beware girls with green knee socks.

Other thing that wasn't mentioned is the actual mechanism that caused that 'controlled randomness':

http://en.wikipedia.org/wiki/V-1_flying_bomb#Guidance_system

http://search.dilbert.com/comic/Random%20Number%20Generator

And of course this one: http://xkcd.com/221/

what does Poisson-distributed randomness look like?
Events generated by Poisson processes or amenable to the small-p binomial approximation look like Poisson. Events not amenable to the small-p approximation look Gaussian. Extreme value measurements (flood water levels, auction prices) look Weibull, Fréchet or Gumbel.

Appropriate statistical methodologies are appropriate. sigh

i read somewhere that showed the V1's had a higher kill rate of RAF pilots than German fighters, with no losses of german pilots. Couldnt find it though, anyone else?

That's extremely unlikely. How is kill rate defined?

I believe he means kill ratio. So, the number of planes shot down versus the number of V-1s shot down. If one RAF fighter was shot down for every two V-1s the ratio would be 2:1 in favor of the RAF.
This is a little bit plausible. This is how the RAF shot down V-1s: http://i.dailymail.co.uk/i/pix/2011/05/08/article-1384740-0B... (picture courtesy of the Daily Fail)
So, the kill ratio of V-1s might have been higher against the RAF because it was so hard to down V-1s and doing so was hazardous. We're still talking a tiny fraction of the craft engaged, though.

Yes, according to the chart above it took down 351 aircraft. The toppling method wasnt the only one implemented, they also tried shooting them down with their guns; however, they would have to get close and often resulted in the destruction of their own aircraft as well.

I wish I'd noticed that chart earlier. You're right, the V1 ratio was much better.

Part of the data are at...
http://en.wikipedia.org/wiki/V-1_flying_bomb#Assessment

Reply on news.ycombinator.com