To cut a long route short

I'll be honest with you. I think I'm a little neurodivergent. I certainly have dyscalculia and might have a tiny streak of OCD in me. The OCD manifests itself when out running. There was a time when I had to absolutely finish a run on a whole mile. I've since expanded it to accept .25, .5 and .75 of a mile, as well as whole kilometres - 5K being 3.1 miles etc:

I have to live with race distances as they are officially measured. But I get a little annoyed when my Garmin reads 26.43 miles on a measured 26.2 route. That's almost a quarter of a mile difference.

I know why it's out. The measured route is recorded with the shortest routes around the corners. The shortest is the fastest. The "racing line" as motor racing drivers call it.

When you're in a big pack of everyday runners, most of the time you're not taking the "racing line". You can't in a mass of people. You have to accept that. 

So I wondered. Just how much can running the racing line affect your distance?

I'm working on an experiment.

A typical American "city block".

This is section of the streets of my home town. I'm going to run the two routes marked. The red around the inside of the block. The blue around the outside, and a longer route. But how much longer is it?

The inside line .35 mile

The outside line .41 mile

The exact rules of such an experiment need ironing out The Garmin should be have some kind of official calibration and I should probably experiment with other GPS's too. It's just an idle idea off the top of my head. But already you can see that my Garmin records that the outside route is .06 of a mile further than the inside. Two laps would equate to a tenth of a mile difference. You can see how you can easily pick up some extra distance if you don't run the official measured route of a race exactly. 

Any thoughts on how to properly conduct such an experiment gratefully received.