Absolutely agree. Although my satnav (Zumo 660) usually shows a positional error of around 4m (lots of hills around here). So in theory it is possible to have a maximum 8m error (4m each way) when fixing two points within a second of each other. At 60mph - 27 m/sec - 8m is a massive percentage error.
But as we both said, given a clear sky and a constant speed and evened out over a few seconds, the error becomes relatively small, and the satnav shows a much greater and usable accuracy. I never said that they could not be accurate. I said that if you take a single, quick glance at the satnav speed, you cannot rely on it to indicate your true speed. The comparison with aircraft isn't quite valid - I believe that commercial aircraft use Doppler for their speed indications, whereas I believe most auto satnavs use distance / time calculations.
I was quite amazed at how accurate some of the really expensive GPS system can be - a magazine over here in the UK strapped one onto the rear of the ST1300 to measure the Pan Weave phenomenon in order to measure the alleged oscillations of the rear end.
No one has yet commented on the fact that the error that you observe on your speedo is not a fixed value. It changes, and is virtually guaranteed to change by at least 2% as the tyre wears down to the minimum tread depth, as well as being affected by all of the other external influences.
