Originally Posted by WFEGb
This is possible. They just need some small acceleration sensors (one for updown and one for forward backward, best a third for left right.) Then the software has to analyse these movements and fit it to the right stroke and count. Should be no big problem to find the actual half strokes, our strokes. Think for now all swim watches just have one of these sensors in and try to predict from cyclic movements and longer pauses (push offs).
Would wish this from the manufacturers, but it seems good enough to sell enough of the actual watches.
Werner, you're probably right when you say it's possible but I doubt it is practical at this point.
I don't think the number of sensors is the problem. With the rise of smart phones, volume has gone up and prices have come down. As far as I know, all those that you mentioned come bundled up in a small package that can easily be integrated in a watch; acceleration sensors and gyroscopic sensors are sufficiently cheap not to be a decisive cost factor in a lifestyle device like the Garmin Swim. I don't know what kinds of sensors the Garmin Swim includes but my guess is that there are more than it can actually make use of.
The problem, I think, lies in the sample rate and--predominantly--the analysis of the data: Not everyone's movements are as clearly distinguishable as those of a world class swimmer. The target group for swim watches mostly consists of triathletes and fitness swimmers. Those usually showcase stroke patterns that are less than optimal and hard to distinguish for such a little device.
You have to take into account that the battery lasts over a year on the Garmin Swim. If you put in the computational power and memory that is required to provide decisions based on heuristics and statistics, which might give you the desired reliability of the analysis, you'd run the battery dry in one session and the CPU heat would burn your arm in spite of the surrounding water. ;)
With limited computational resources and the need to work on the least common denominator in terms of stroke mechanics, I think the manufacturers are doing a pretty good job. They're not lazy, stupid, or greedy; the required low power technology simply isn't available yet at a price anyone but the military would want to pay.
Give it 10 or 15 more years and the chips implanted into your forearms, powered by your movement, will provide all that data live on your contact lense display while you're swimming. :)