Difference between revisions of "Talk:2009-04-24 48 11"
From Geohashing
imported>Ekorren (Birthday paradox.) |
imported>Ekorren (Very high chance of near hashes.) |
||
| Line 5: | Line 5: | ||
[http://irc.peeron.com/xkcd/map/map.html?date=2009-04-24&lat=48&long=11&zoom=9&abs=-1 Map of 2009-04-23] | [http://irc.peeron.com/xkcd/map/map.html?date=2009-04-24&lat=48&long=11&zoom=9&abs=-1 Map of 2009-04-23] | ||
--[[User:Zb|Zb]] 20:07, 23 April 2009 (UTC) | --[[User:Zb|Zb]] 20:07, 23 April 2009 (UTC) | ||
| − | :I believe noone has exactly examined that yet, but basically those repetitions seem to be an application of the [http://en.wikipedia.org/wiki/Birthday_Paradox Birthday Paradox]. In our latitudes, the chance is about 3.8%, i.e. once in four weeks in average, that two consecutive coordinates are within of 10 km of another. | + | :I believe noone has exactly examined that yet, but basically those repetitions seem to be an application of the [http://en.wikipedia.org/wiki/Birthday_Paradox Birthday Paradox]. In our latitudes, the chance is about 3.8%, i.e. once in four weeks in average, that two consecutive coordinates are within of 10 km of another. |
| + | If I haven't calculated total crap (my stochastical knowledge is extremly rusty, there is hardly any iron left), the theoretical chance that within of a given week (i.e. one calendar week) two arbitrary hashes fall within of ten km of another is about 59% in a latitude 48 graticule. That means, such an event is all but uncommon.--[[User:Ekorren|Ekorren]] 22:11, 23 April 2009 (UTC) | ||
Revision as of 22:11, 23 April 2009
Flippin' weird: Is it just me, or is Mammendorf is just as preferred by the random (?) Geohash algorithm as the surroundings of Munich's airport? 2008-07-23 48 11 2009-04-03 48 11 Map of 2009-04-23 Map of 2009-04-23 --Zb 20:07, 23 April 2009 (UTC)
- I believe noone has exactly examined that yet, but basically those repetitions seem to be an application of the Birthday Paradox. In our latitudes, the chance is about 3.8%, i.e. once in four weeks in average, that two consecutive coordinates are within of 10 km of another.
If I haven't calculated total crap (my stochastical knowledge is extremly rusty, there is hardly any iron left), the theoretical chance that within of a given week (i.e. one calendar week) two arbitrary hashes fall within of ten km of another is about 59% in a latitude 48 graticule. That means, such an event is all but uncommon.--Ekorren 22:11, 23 April 2009 (UTC)
