OpenStreetMap
Well, apparently I'm in the top 50 (of all time?) of OSM users for uploads of GPS data: http://www.openstreetmap.org/stats/data_stats.html .
That's nuts. I don't even track half of my car/bike/walking trips. You know what this really means? It means the *rest of you* aren't uploading enough GPS data. :)
So, go on, get to work! Trace *all* your trips. Multiple traces on the same road are useful for statistic analysis and traffic modelling, so you can't use that as an excuse either. :)
Discussion
Comment from Anders Lund on 20 January 2009 at 09:01
Yeah ... I know ... I must do more... sorry.... :-)
Comment from smsm1 on 20 January 2009 at 12:09
I've not got around to uploading any traces in a while. Really need a sync utility, so that they are uploaded automatically.
Comment from RichardB on 20 January 2009 at 15:17
You've got no excuse, smsm1 - you can trace 5-times quicker than me...
David, you say "of all time?" - yes, it is an all-time list - however you'll notice number 1 on the list is "Lost GPX traces" - for some reason at the back end of 2007 a number of GPX traces were mysteriously wiped from the database (including my first ever trace from September 2007). They were all recovered - but without any data about who they came from - so they were all placed in this one user account. You could get them back if you knew what they were called or what tags they had.
i.e. some people may have actually uploaded more than the database currently says.
You say multiple traces are useful for traffic modelling - hmm, if they did that in half the areas I've mapped then they'll believe that all users turn left every time they get to a side turning and they stop randomly at post-boxes, pubs, churches, schools, etc.
Having said that, multiple traces, particularly taken on different days, are useful for "averaging out" position errors in individual traces - and I'm sure in future the likes of JOSM will be able to do some averaging by looking at the density of traces.
On the other hand, my residential cul-de-sac probably has a higher density of traces than the nearest 8-lane motorway a few miles away...
Comment from Richard on 20 January 2009 at 17:25
I guess the other thing is that logging at 1pt/sec will propel you up the leaderboard. I rarely go that fine - there's not much point if you're cycling along a country road - and in fact my first few years' traces were done on an old yellow eTrex which only did the 'auto' logging anyway (just 1000 trackpoints, and more logging when you go round corners). Far too computationally expensive, of course, but I'd love to see a ranking based on distance covered...
Comment from David Dean on 20 January 2009 at 23:33
RichardB, with regards to the possibilities of traffic modelling, that is why tracing every trip would be useful, not just the ones in which you are actually doing OSM mapping.
Comment from walley on 21 January 2009 at 20:43
I have 330mb of tracks, i'll upload them someday, I hope!:) (2177289 points)
Comment from RichardB on 21 January 2009 at 21:05
DavidDean: Yes of course, but there are some roads that no OSM-ers will otherwise travel on for many years in some cases, apart from when they're doing systematic mapping. Even though there are quite a lot of us these days, there still aren't *that* many in the grand scheme of things. These roads would give very odd results when you tried to do any kind of analysis to them. In addition, roads that are travelled by others will be tainted with this odd data, which might be difficult to tell it apart from heavy traffic in places.
I doubt we'll get to the stage that we'll be able to do much useful traffic analysis with this data anytime soon - especially as much of it has been done on foot or by bike (how do you know if one has been done by bike or has been done in a car in relatively heavy traffic?).
If I'm doing a long-distance journey somewhere - or a journey to somewhere more local that I know hasn't got many traces - then I'll definitely put the GPS in the car, but I don't record my daily trip to work and back each day (but of course I've done it once or twice).
If we want to do traffic analysis - then we need lots of data covering all days of the week, all times of the day, and the entire road network ideally. This might be better as a separate (but connected) project?