I'm noticing that there's about a three percent a discrepancy between what my bicycle computer reports and what the GPS-based apps in my phone report for the same ride.  Example: yesterday my bike computer told me I rode 29.67 miles but Ride With GPS says it covered 28.8.

Since I can measure the circumference of my front tire to within a couple mm and enter that number as the wheel size setting in the bike computer, I'm inclined to believe the bike computer's number is closer to actual than what the GPS-based numbers tell me.

What it means in practical terms is that I can use the RWGPS figures as a planning estimate but should not expect to be arriving at my destination when the bike computer reaches the number provided by RWGPS: I'll have a few more miles to go.  At the end of a long hard day that can be very discouraging.

Anyone else notice a similar discrepancy, and suffer similar disappointments?

I can’t say I’m disappointed, but I’ve long noticed that the distance my computer says I’ve ridden can differ noticeably (more than a km) from what my riding buddy’s says about their ride, even though we started and ended within a couple of metres of each other. Much more mystifying, though, is the difference in elevation change recorded. But then, the elevation accuracy of most cycling GPS devices isn’t what we’d expect. 

I can’t speak to differences between different device types, but Rachael and I have always ended up with different stats on ours at the end of the ride, even though they’re the same device and when we’ve ridden essentially identical rides.  

It doesn’t bother us though; and of course we always assume the one with the highest reading is the correct one and take credit for it.

Ah yes, the mysterious "transporter effect": you must have been momentarily teleported and not realized it.

re: elevation errors.  It always seems to work out that on a lop ride the total ascending is not equal to the total descending, although I seldom notice any sort of actual "step" one way or the other when I dismount.

Did you know that there is an actual difference in "sea level" between the Atlantic and Pacific Oceans?  It's something like six feet; it's a wonder that shipping doesn't crash into it.

Well of course max distance wins!  :)

It's not surprising that the measured distance between two (or more) riders will vary slightly. Being based on the wheel size, unless you've actually measured the size of the wheel there's built-in error.  (Even then, there is probably some error based on changes in tire pressure over time; if one wheel loses air at a different rate than the other, the size will change accordingly and throw off the computed distance).

I've learned over the years never to use the numbers printed in the cycle computer instructions; I think they must be based on either an "average" or a calculated hypothetical wheel circumference.  They never match what is actually on my bike.  

I always measure (roll the wheel along my tape measure using the valve stem as the marker to determine one full revolution) the circumference that is yielded by my particular combination of rim and tire selection and enter that number into the computer.  I don't know whether, or how I would go about determining by how much, to compensate for the slight compression of the tire when in use under load.  Some things just aren't important enough to worry about.

Oh yeah, one other thing about (my) Garmin: when I am following a stored route, it periodically loses its mind in the middle of a ride and tells me I have several thousand miles to go before the next turn.  It's not consistent, and sometimes self-corrects later on during the ride but not always.  But it always correctly shows my true location with regard to the purple route line, which is mostly what I pay attention to anyhow.

I usually have the turn-by-turn navigation turned off and just the line of the course visible. I think newer devices than mine are better, but I haven’t got one yet. 

As I understand it, the devices at the price level of cycling computers (and smartphones) use atmospheric pressure to determine elevation changes. As the weather changes through the day, you may find your loop ride ends at a higher or lower elevation than it started.  My day ride to Île de Ré shows this quite well:

https://www.cycleblaze.com/journals/france2019/ile-de-re/#1579714571584

I had no idea that a phone or consumer-level GPS would have a barometric pressure sensor for determining elevation but the observed behavior certainly fits the idea.

My Master's degree project involved using a barometric altimeter (this was 38 years ago, long before smart phones).  We would measure the altitude at a given location, work for a couple hours, then return to the original location and measure again.  Any difference between the measurements was divided and distributed proportionately between everything we did in between the two "baseline" measurements.

We periodically also visited positions whose elevations had been determined by survey methods, to give us a correspondence between the instrument elevation and actual.  That discrepancy was also factored in when making elevation measurements elsewhere.

Incidentally I've covered that same approximate route!  It was the first day's ride of a two-week (supported, pre-organized) tandem group tour that began and ended in La Rochelle, ten (gasp!) years ago.

We had a flat, about in the middle of the bridge.  :(

Photos at my Google Photos album from our 2012 trip to France if you want to look at someone else's vacation pictures.  :)

That's also, if I remember correctly, where the donkeys wear "pants" to help protect them from the nasty biting flies that inhabit the tidal flats where sea salt is produced.