The ka72 site had another update last night. The new "Big List" page allows you to do more sophisticated searching and sorting of tracks. I also added a field for Track Type, as I am starting to see more tracks from other than windsurfing.

The most users come to the site from Australia, followed by the Netherlands, then the USA is in third place.

Looks like (allowing for the weirdness of a couple of files) the fastest places I've had tracks from so far are (in order):
Lyle, Washington, USA (42.23kts on 30 August 2008)
Melville Beach, WA (41.98 kts on 19 February 2009)
Hindmarsh Island, SA (41.84 kts on 8 November 2008)
Hindmarsh Island, SA (41.47 kts on 17 May 2008)
Chinook, Washington, USA (41.12 kts on 4th october 2008)
Mandurah, WA (40.95 kts on 30 December 2007)
Burrum Heads, Qld (40.92 kts on 9 March 2009)
Mandurah, WA (40.67 kts on 6 February 2009)

www.ka72.com/

Dylan.

That's amazing Dylan, it's incredible how quickly news spreads on the grapevine.

Is it worth including an invite on your site, for teams to join the challenge?
I'm sure Nebs would love to have teams from the Netherlands and the USA.

Yes, well, if Nebs gets off his butt and sends me a banner+link to put on the site, I'm happy to oblige.

By the way, Vando has kindly uploaded a track to the site tonight that puts Burrum Heads firmly in first place.

Select to expand quote

Dylan72 said...

Yes, well, if Nebs gets off his butt and sends me a banner+link to put on the site, I'm happy to oblige.

By the way, Vando has kindly uploaded a track to the site tonight that puts Burrum Heads firmly in first place.

Yes well we couldnt have those USA guys in front could we

Dylan great job mate too easy

Select to expand quote

Dylan72 said...

Yes, well, if Nebs gets off his butt and sends me a banner+link to put on the site, I'm happy to oblige.

By the way, Vando has kindly uploaded a track to the site tonight that puts Burrum Heads firmly in first place.

Does the site pass the six test cases I sent you yet?

Good answer nebbian.I wonder? Hope so.

EXCELLENT - I just got a 5 knot improvement in my PB by uploading my sbn file to ka72 - and it was an ocean sail where realspeed robbed me and only gave me 25K top speed.

But to be fair I dont even remember going to Kalgoorlie that day ! (checkout the google overlay)

(in realspeed I always have to remove invalid points, some of which may be to do with me being completely underwater or upside down) I did bail out of a few jumps and lose it badly on a wave that day so it wasnt a real sensible speed session but maybe the software could look for those impossible accelerations and remove those points. Just an idea.

I dont know if it is a problem with my gps but I suspect other people may have similar issues. It got a lot better once I started turning on the GPS first and then rebooting it before going sailing.

---

I just uploaded the cleaned file and I think the numbers it gives agree with realspeed, just noticed I had shortchanged myself on distance sailed that day (had a flat laptop battery and had save the results to csv so had them but "remembered" the distance.)

Select to expand quote

nebbian said...

Does the site pass the six test cases I sent you yet?

Let's forget for a moment that you have misunderstood the conversation.

What you are asking is not exactly a fair question.

There isn't a software package on the market that gives the results you expect for all those files.

The files you supplied mostly require human analysis to fix or alter them. That is something that requires a human. The ka72 site allows you to inspect files after they are posted, which is something that no other package provides. When someone posts results from Realspeed (for instance), you have to request the file from them if you suspect an inaccurate result. With ka72.com the file is there for you or anyone else to check. The site reduces human error by taking control of the analysis settings away from the human, and then enables human inspection to fix those mechanical errors that require human intervention.

We should really take this offline, but I think it's worth saying that IMO if you expect the ka72 reader to do your job for you, you totally miss the point.

Now, my original note was about whether the GPSTC would like to be linked to from the ka72 site, which is entirely welcome any time, and has nothing to do with whether the analysis engine passes any tests. I would encourage any gps comp organiser that is interested in being linked to from the ka72.com site to send me a banner and link, and I'll post it gratis. Same goes for the analysis software sites.

My goal is to make it easier for people to start on the speed-sailing journey. The site isn't a crutch for competition organisers, and it isn't meant to replace the human side of track auditing.

If any competition organiser has a set of files they want to use to audit the site, I am happy to tackle that task if it means that people out there can use the site "officially" for their comps, but the site is not going to produce any different results than GPSResults or Realspeed would when you give them garbage files. It is not fair to expect the site to produce results that are only obtainable normally by manually massaging the data or other manual tricks.

(Having said that, there is one thing that the engine is capable of, but which the site doesn't provide, and that is the ability to combine multiple files into a single session. This isn't a code limitation, it is simply a user-interface issue that I haven't tackled yet. One of Nebbian's tests involved manually combining two files. I still think this is a manual process, as there is no way for the computer to *know* that two files are from the same session.)

Select to expand quote

hardpole said...
EXCELLENT - I just got a 5 knot improvement in my PB by uploading my sbn file to ka72 - and it was an ocean sail where realspeed robbed me and only gave me 25K top speed.

This, on the other hand, is a fair test. If Realspeed can mark a point as invalid, then the ka72.com site should be able to as well. This kind of error is worth investigating.

If the test files provided by Nebs were more like this, then I think we'd be getting somewhere.

Select to expand quote

hardpole said...


I just uploaded the cleaned file and I think the numbers it gives agree with realspeed, just noticed I had shortchanged myself on distance sailed that day (had a flat laptop battery and had save the results to csv so had them but "remembered" the distance.)

Another potential issue with this file, apart from the spikes, is that it includes a significant amount of driving. This can have a big effect on your top speed! Of course, if you MANUALLY cut out the driving, you are ok. But that is exactly the sort of manual operation the site isn't capable of doing for you.

Realspeed, once the spikes are removed (via "remove invalid points") gives a 2 second peak of 36.4 knots, exactly like ka72.com does. The problem is that the peak was obtained during the drive home.

OOPS - sorry Dylan - forgot about that -the points near coolgardie threw me off. Your right I did drive home with it on.

I dont know which emoticon to put for embarressment !!!

I was looking for one of my files that have spikes to see what would happen and as luck would have it the one I picked also had the drive home in it. It sounds like the site has the logic of spike handling working.

sorry to waste your time, I think thats the second time I have done that. I should just keep quiet.

I did post before, and now it sounds like crawling, that the google plugin thing is very cool. Nice for quickly showing people where you have been etc...

And even worse - now you know where I live !

Select to expand quote

hardpole said...

And even worse - now you know where I live !

Track can be pulled if you are concerned.

No worries, Hardie. I hope I didn't come off sounding too sh***y above either. Been a long week. Honestly, I hope the site is useful. Those who know me well know that (1) I often appear more brusque than I mean to and (2) I really don't know as much about windsurfing as I make out.

You guys do all the hard work, and I just try to make it easier for some of you.

I take it that's a "no" then Dylan?

When the KA72 site passes those test cases (I don't mind if there is some manual intervention, for example to combine files) then I'll be happy to endorse it with a banner, and hopefully set things up so that it has easy integration into the GPS Team Challenge website.

Select to expand quote

nebbian said...

I take it that's a "no" then Dylan?

Nice try.

The site gives the same results (within about 2%) as Realspeed and/or GPSResults for all the files that you provided.

What it doesn't always give is the results you wanted, which are also not given by the offline software.

If the test is to reproduce the results of the offline packages, it passes.