The New FON Statistics
It’s Sunday, and I had nothing better to do this morning besides downloading all of the separate database files from the POI tab at maps.fon.com and do some simple statistics. There are 90 countries represented, with a total of 52,089 52,088 Fon hotspots, including exactly 200 which are apparently located at the North Pole.
The top 34 countries (from POI files >1k), in order from most hotspots to least are:
01 Germany: 11039
02 SKorea: 7977
03 Spain: 6339
04 France: 3464
05 USA: 3463
06 Japan: 3220
07 Sweden: 2965
08 Italy: 2625
09 Taiwan: 2019
10 HongKong: 1693
11 Netherlands: 1327
12 UK: 1039
13 Austria: 986
14 Finland: 882
15 China: 688
16 Denmark: 579
17 Belgium: 346
18 Portugal: 308
19 Hungary: 164
20 Ireland: 145
21 NKorea: 88
22 Norway: 85
23 China: 83
24 Greece: 69
25 Poland: 64
26 Argentina: 52
27 Slovenia: 43
28 Estonia: 32
29 Canada: 32
30 Brazil: 26
31 CzechRep: 22
32 Luxembourg: 19
33 Mexico: 17
34 Australia: 16
It would be wrong to conclude that between Sept 25, 2006 and Jan 21, 2007, Fon has grown by approximately 48,400 hotspots, as the POI function is undoubtedly (and understandably) including hotspots that have not been active within the hour of downloading. The exported files would not be used right away, perhaps days later after flying overseas. Mike Puchol made an effort to eliminate inactive hotspots during his original investigation (just before the POI tab was yanked away the first time).
I suspect that Fon does not update these coordinates in real time, if only because there have been absolutely no additions in the USA in the past 3 days. It appears that adjustments in the hotspot’s position on the map are updated almost immediately. Changes in my address text appeared after 15 mins or so, but it took over a half hour for the star to re-appear on my hotspot after I turned it back on again. If there isn’t an instant-update system, then at least it’s being done regularly.
Please regard the rankings higher on the list to be more accurate than the ones at the very bottom. Since I eliminated 2/3 of the databases just because they were smaller than 1 kilobyte, there could be some countries that were included or eliminated only because of the average text length of street addresses there.
I was a little suprised at the results. Overall, I thought there would be more hotspots everywhere. In mid-December, Fon claimed to have more than 100,000 hotspots worldwide, twice the number documented in the POI, of which, perhaps 4 out of 5 have not been on “in the last hour”. Also, the last I had heard, the USA was a very close second behind Spain, who was in the lead. It would appear that Germany is the clear winner now! I was also suprised to see how few hotspots were with the USA’s neighbors: Canada and Mexico. I can understand about Canada due to their smaller population (and Fon does not ship there), but Mexico has many high-density, modern cities, and should not suffer from the still-aggravating language barrier.
By adding together the 5 countries represented above, which predominately speak English (USA, UK, Ireland, Canada, Australia), I get a count of 4695 which puts us in 4th place worldwide. This should mean something to Fon, which still hasn’t assigned a permanent moderator to the English discussion board who properly reads and writes the language. I know it seems mean of me to say this, except that his poor English has resulted in numerous conflicts and forced others to follow up on his support answers to explain what he meant. Speaking of discussion boards, it looks like Fon is WAY overdue in launching an official board for Germany, and perhaps Japan next.
In conclusion, I concede that the maps export feature is not a perfectly accurate method of determining the population of Fon routers being operated. While Fon may declare that there are hotspots everywhere, all that we know comes from these statistics that we pull from them like pulling teeth. These are undeniably overstated. If anyone has done an up-to-date survey of ACTIVE hotspots worldwide, I would love to know. 
January 21st, 2007 at 12:40 pm
About growth:
It will be very interesting to see if FON will claim the maximum growth based on POI extraction.
The measurements that were performed in September, FON always considered way too low, inaccurate, and FON claimed to have a lot more usable hotspots than the available figures indicated.
Now, if that figure was too low, and by re-instating the POI tab FON recognizes that the current extraction is correct (although no longer discriminable on activity or “flag”-only entries), the growth must be lower than we would expect by the actual numbers…..Right ?
So, in future discussions, we will talk about growth of (as I compute) (much) less than 41.000 usable hotspots in 4 months. (sorry FON, you can’t have it both ways).
Lets say 10.000 each month.
What I’m curious about is how much of this growth is solely caused by giving Fonera’s away for (almost) free. What part of the growth is statistically irrelevant, and has no weight in the prediction of future growth at all?
FON will most probably not reach previously predicted goals this year.
It can’t afford to stop the give-away policy and at the same time will have to invest in all other things that were blatantly ignored the last year, like supporting the people that run the hotspots for FON.
So far FON has not shown the necessary capabilities to combine these, and other, tasks.
January 21st, 2007 at 1:56 pm
Keep in mind that Mike Puchol did not present the statistics from the combined POI files as I have. Mike interfaced directly with the map engine in order to request only “active” routers. The POI tab has probably always given an exaggerated count. I’m only attempting to document them at their current numbers, since it’s as close to an official number we’ll ever get. Fon has changed the accessability mechanism of the hotspot database in order to prevent anyone from downloading the entire list in any reasonable amount of time. They can still claim that the new POI databases are also undercounts, for “technical” reasons.
January 21st, 2007 at 4:43 pm
You beat me to it! Very interesting figures, it would be interesting to compare a .kml export with the official maps page, to see if the ‘light green’ icons are also present. That would give you a better idea about the ratio of possibly online to really online. What didn’t make it into my analysis (basically because they removed the possibility to do so) was a running snapshot, that actually looked at the variance every few hours.
In my post, I saw the highest router ID to be around 20.000, this would be the figure of ‘have-been-seen-at-least-once’ routers, if they had all been registered. If they have spent just under 10 million Euro (source: Martin’s blog), we can conclude that the COA for each running hotspot is about 200 Euro. How many passes need to be sold to make each one break even…well you do the math
January 22nd, 2007 at 1:14 pm
Mike / Austin:
in an ov2 file of the Netherlands, light green hotspots are included.
Orange (inactive members) are not included.
I suppose the result of a bit more than 41000 hotspots growth (addition of light and darkgreen spots) is correct. (10000/month).
January 22nd, 2007 at 1:17 pm
aaaaah: 10,000 / year that is…..
January 22nd, 2007 at 1:32 pm
I assumed that the orange dots were registered Foneros who are potential “Aliens”, dark green were locations of routers which have been down more than an hour (including ones down weeks or months), and the bright fuzzy green blobs, which are about 5 times larger than the effective range, marks a hotspot that has reported in to the mothership within the last hour. If that’s even it’s actual location anymore.
January 23rd, 2007 at 5:18 am
Not quite:
- orange: Foneros without registered routers (according to Martin, potential Aliens, but I think originally any Fonero was registered orange, including people that bought but didn’t register a router)
- light green: registered routers that didn’t show activity in the last hour
- dark green: registered routers that did show activity in the last hour
- dark green with white star: dark green AND personalized page.
That is information from the map site, check the config options.
The information given, is incorrect (of course…).
At my address, two routers are portrayed, one light green (a non-existing router), one darkgreen with whitestar.
The last is an existing FON router, where nobody logged on the last week. So active is apparently equal to the thinclient being active… and doesn’t depend on user activity.
And that combined, would lead to the conclusion that all light-green icons are not alive and kicking; if the thinclients would work, they should be dark green. (Or am I jumping to conclusions ?)
January 23rd, 2007 at 8:29 am
Euronerd: thanks for the explanation. I think the map should contain a legend explaining the different colors because they are confusing (of course). I think of light as being on, and dark as being off. Perhaps I am unusual?
Also, I cannot see the so-called “personalizations” of hotspots that have been down an hour or more. The star only appears when the hotspot is currently up. How can I find out who it is, to call them and ask them to reboot their router?
It was a mistake to use green to mark hotspots. Green is usually difficult to see on the maps (in hybrid mode) due to vegetation. Also, green is not a color that makes me think of wifi. It was a mistake to use a solid orange dot for registered Foneros. The dot makes the map underneath unreadable. Orange is Fon’s primary color; it should be the color used to mark hotspots.
The icon for a hotspot should be an orange circle outline with a tranlucent middle so that the map underneath is still visible. If the hotspot is down, the outline should become dotted, and the middle transparent. Spots down more than a couple days should turn grey and be removed after a week. The hotspot sizes should be half the size as currently shown on the map. The current green balls cover a whole city block. Come on, now. I may be able to see an SSID from almost that far, but the useful range is half, or less of that.
I understand Fon is turning inactive “Linuses” into “Aliens” now, so lots of those orange dots may be unsucessful routers. Too bad Fon doesn’t host a Fonero email-forwarding service so I could try to contact these guys and visit them in person to set their routers up.
Aliens should be given the choice to not show their locations. Privacy is a concern to many people. I don’t want to be one of the victims of “The FON Massacre” when some psycho uses these maps to track down “Aliens” and murders them!
The icon for an Alien should be a little stick-figure person with a transparent background. Alien icons should be invisible by default and hotspot owners should have some process to have Alien icons removed (in case 100 Aliens register with a shop’s address as theirs - you wouldn’t be able to see the hotspot with all those icons piled up!).
January 23rd, 2007 at 5:54 pm
North Korea is given to host 88 FON access points ? Sounds like a misleading news.