SportTracks versus Ascent

200904300356.jpg 48 hours ago I couldn’t have run round a goldfish bowl. I’m happy to say now that everything feels like it’s back to normal and I can even walk down stairs without wincing at every step. That feels like real progress.

For as long as I have been using GPS devices to track my running (I started with the Forerunner 201 and am now on my third 305, having ditched the 405 after a single run), I have used SportTracks from ZoneFiveSoftware. It’s so good I’ve paid for it twice. And, I even run it on a PC that sits on my desk as I haven’t managed to get it to work reliably on my Mac, despite trying Parallels and Fusion. Others seem to have managed it, yet every time I’ve tried it, I’ve had trouble importing my history file, trouble getting it to reopen a logbook after exiting the application, trouble creating new logbooks and so on. So I gave up and stuck to the Vaio laptop on my desk that, once a week, would get opened, booted and used only for SportTracks.

I don’t have to do that anymore. There’s a new kid in town. Well, new to me. It’s called Ascent, from Monte Bello software. It sells for £29, although a trial download lets you get used to it with up to 10 activities. So far, it rocks. It’s as good as SportTracks – and I think that’s saying something as SportTracks is really good. Ascent imported all of my old data going back to 2005 without even blinking – and even recognised the duplicates from the trial install where I’d synced it with my 305 and its most recent runs.

Here are a couple of data items from Ascent, from Sunday’s London Marathon.

1) First, my pace graph. The start was very slow – it was more stop than start in fact. A bunch of guys were running out of pen 2 (runners aiming for 3:30 or better), roped together in a large square. I don’t know what time they ran but I know people who ran 4:45 who passed them so I’m sure it wasn’t 3:30 – but where the road narrowed early on, and there was a dense crush of people trying to get through, these guys stopped everyone in their tracks. Still, that might have been a good thing – fast starts are supposed to be bad after all. After that, I stayed pretty consisent, varying +/- 10 seconds from my target of around 5m 20s per km all the way up until about km 22. From then on, it all went wrong until km 35 when I gritted my teeth and tried to get home in at least a reasonable time

200904300404.jpg

2) And then my 5km split times – it’s pretty easy to see how much I slowed down from km 20 onwards – really suffering from 30-35.

200904300410.jpg

Pace graphs in SportTracks are slightly clearer perhaps – but that’s because it doesn’t start the y-axis at zero and so the differences are accentuated. I need to figure out how to do that in Ascent or perhaps request it as an enhancement in the next version. It’s quite clear from the graph below where I had to stop and walk for a bit! Looking at these two graphs more closely, along with the accompanying split data seems to show some differences in recorded times per km (some are 2 or 3 seconds, which would be accounted for by GPS smoothing and others are longer – e.g. km 28 on SportTracks is 7:58 yet is 6:27 in Ascent. I need to do some work on that.

200904300444.jpg

I like this new software. It will be great not to have to boot the PC – and I’ll miss SportTracks but all good things eventually get replaced. Well, nearly all.

Crowd Sourcing London Marathon Photos

200904282145.jpgBrowsing through some photos from the marathon, I wondered why the folks that take pictures at marathons around the world and then sell them to you don’t crowd source their work.

There were maybe 250,000 or even 300,000 people out watching the London Marathon this weekend and many of them had cameras.

Assuming that the same technology that google uses to blur faces and number plates in Street View and the same technology that captures number plates for the Congestion Charge can be used to read the numbers/names on photos such as the one at left …

Then why couldn’t anyone who took photos upload them to a site where they could be searched by anyone looking for photos of themselves running?

A download would result in a small payment to the person who took the photo and a payment to the marathon photo company for facilitating the transaction. It would be like the iPhone App Store, but you wouldn’t need any coding talent.

I’m pretty sure all of this could be pretty much fully automated and so the variety of photos available to runners would be multiplied enormously at practically no extra cost to the services that allow you to search for photos.

Economical Forecasting

The weather forecast for tomorrow has shifted again

200904251143.jpg

If this holds, then the rain has shifted to Monday and the weather for tomorrow’s marathon looks “ok” – not great, but ok.

But, I did wonder today whether both of these paragraphs are true:

The weather is a complex system for which there are many variables. It takes a great deal of computing power to figure out even very localised immediate forecasts, let alone system wide forecasts. And forecasts – even at a trend level (i.e. the next few days will see more sun than rain) – are notoriously inaccurate. Such predictions frequently get duration and severity wrong and their results are often the exact opposite of what actually happens. Forecasters will hate the use of that word “prediction” which implies crystal ball gazing rather than the use of multi-faceted models built by the brightest people in the land backed by years of research. But the empirical data says they’re wrong more often that they’re right.

The economy is a complex system for which there are many variables. It takes a great deal of computing power to figure out even very localised immediate forecasts, let alone system wide forecasts. And forecasts – even at a trend level (e.g. house prices will next rise, rather than fall) – are notoriously inaccurate. Such predictions frequently get duration and severity wrong and their results are often the exact opposite of what actually happens. Forecasters will hate the use of that word “prediction” which implies crystal ball gazing rather than the use of multi-faceted models built by the brightest people in the land backed by years of research. But the empirical data says they’re wrong more often that they’re right.

No wonder we’re not sure if the economy is coming out of recession, going deeper, stabilising, growing or shrinking. House prices up? Down? Flip a coin? Build a model? Play Baccarat as a cipher for prediction?