[OSM-talk] Raw GPS layer

[Mark Williams]

80n wrote:
> On Fri, Feb 22, 2008 at 8:23 AM, Mark Williams <mark.666 at blueyonder.co.uk>
> wrote:
> 
>> Stephen Hope wrote:
>>> This would be good.  But even better, let me select a portion of a
>>> track log and upload it.  My track logs tend to be a nightmarish
>>> tangle, with possibly hours of stuff before, after and during the
>>> interesting bits.  I can use them because I was there, and know where
>>> I went, when and why (this is why I take notes).  But somebody looking
>>> at the raw track would actually be confusing, and possibly wrong.
>>>
>>> However - bits that I'm actually mapping tend to be much better -
>>> actually tracking roads, paths etc.  If I could easily select the bad
>>> bits of the track log (just points) in JOSM and remove them, then
>>> upload the rest, I'd be willing to put them up.
>>>
>>> I keep meaning to go back over my old track logs (all of which I have)
>>> and clean them up with some 3rd party tool, bit I always seem to have
>>> new stuff to work on instead.
>>>
>>> On 22/02/2008, Frederik Ramm <frederik at remote.org> wrote:
>>>>  I think we should provide a track upload facility within JOSM. I
>>>>  started work on that once but got distracted, maybe its time to
>>>>  revisit that.
>>>>
>>> Stephen
>> One other point is that a track layer will highlight all our homes in a
>> very public way; at present you have to download something eg josm + GPX
>> trackdata to see this at a meaningful scale (ie not Potlatch, for this
>> purpose). This effectively reduces the casual browsers chance of
>> noticing the possibility, but posting it publicly hangs out a banner.
>>
>> I am aware of at least one user with a node marking his home, so we
>> don't all care, but it's worth considering first!
>>
>> Also, I don't really see the utility of this, even after reading the
>> preceding posts. You can't use the data from a visual map of traces for
>> much, and areas where doubt exists eg changes to roads, will have a mass
>> of new & old to make a mess there...
>>
> 
> 1) They prove the source of your contribution, in the same way that a good
> Wikipedia article cites its sources.  Several of the reasons listed here:
> http://en.wikipedia.org/wiki/Wikipedia:Citing_sources#Why_sources_should_be_citedare
> equally applicable to OSM.
> 
> 2) Track logs from multiple sources are aggregated.  Different users, at
> different times, and using different equipment will result in a much better
> dataset than a single track log ever can.  It is very common for parts of a
> single track to be off by a considerable amount, this type of error can be
> reduced and eliminated if there are multiple tracks to refer to. If you
> download the tracks for a part of the M25 motorway, for example, you will
> see that the aggregated result is much better than any one single track.
> You'll also notice outlier tracks which can easily be discounted.
> 
> 3) There may be uses of the track logs in the future that have not yet been
> developed or thought of.  For example, it might be that detection of edits
> in places that are distant from any track log could help to monitor for
> vandalism, or indicate a higher priority for peer review.  Analysis of
> average speed and direction might help routing software to determine journey
> times and one-ways streets. etc. etc.
> 
> You raise the point about some of your tracklogs being a bit of a mess.  In
> my opinion you can and should still upload them.  Any analysis of tracks
> will have to use statistical techniques to filter out noise, so anomalies
> will get removed as part of this process.  In fact, many years from now,
> historians and archaeologists will be horrified that our enormous archive of
> GPS data was so badly mutilated before it was uploaded.
> 
> 80n
> 
> 

I wasn't saying not to upload them - just that I'm personally not that
keen to see a raw GPS track layer on the map. I do upload them, that's
why it would show up...

Mark