I don't think that’s the issue. The issue is comfort and ease of use. Race style gear just isn’t comfortable when driving.

A tip for cutting seatbelt. Cut at a angle, not 90 degree. Not sure how to explain, but when cutting at 90 degree, the belts seems to fold up and get caught in hook/safety type seatbelt cutters. But cutting a seatbelt is fast and easy in a 45 degree angle

You said a mouthful. I've ridden in lots of rigs with racing harnesses, they are a pain to put on. But the difference in a rollover is night and day.

But for pavement driving I'll be leaving the regular belts in also when I put harnesses in my Jeep.

Fail SAFE.

If you can cut yourself away, you are aware enough of your enviroment to perform a primary assessment of your injuries, the location and stability of the vehicle, and can make some reasonable fine motor motions. On the other hand, if you are knocked out by banging your head on the window during an accident that includes a roll over, the last thing you want is to have your seatbelt release on you, letting you get tossed about. Heck, you wouldn't that if you were concussed, but if my bell has been rung I'd rather be dangling like a pinata than being a lump of rapidly cooling spare parts. (There is very little of Vermont that is flat, if you leave the road there is a very real chance you are suddenly going down hill in an axis of travel other than the designers hoped.)

Never mind that if it the seat belt releases under a shock load of a couple of tons, you've just left the windshield during a 40mph collision.

I want that sucker to jam up good and tight

Sorry, I just don't think it's rocket surgery to design a buckle that:

#1 won't fail open at design load
and
#2 will reliably release after being subject to design load

What I have heard a few times was the location of the buckle was a problem. It was easier to cut the seatbelt than it was to get past the person and shift them enough to reach the buckle.
I have been told the problem is worse with "Large" people because they are harder to reach around when they are in a car.

I wouldn't be so sure.

Besides, you left out three further parameters:
1, instinctive understanding of mechanism
2, positive function after potentially tens of thousands of usages
3, cheap

You left out the roll cage... There is welded a virtually indestructible cage of steel around the volume which the driver + map reader occupies. With that kind of protection + the real harness you don't need air bags. (And the way these guys drive, there is a very real risk that the air bag could be set off during hard landings).

I'd like to cover a few issues in this thread.

First of all, like some of the other folks here who were/are first responders, we prefer crashes with people in seatbelts, because that usually means extrication, not recovery.

To address some other points:

Jamming: seatbelts come in a range of engineering concepts, all share some sort of intertial sensing that locks the belt on impact (it's a pawl on a ratchet sort of assembly) and the force of impact can cause deformation in the structure of the belt latch, so that's why they sometimes jam.

Airbag seatbelts: all airbag systems deflate moments after deployment. in the case of the ford seatbelt airbags, the same concept applies. the entire belt is not an airbag, so there are cut points for ordinary tools (I used Medic shears almost exclusively to cut seatbelts on extrications. Medic shears are advanced alien technology).

Fixed (racing) vs. adjustable: a 5+ point harness is a great thing - IF you are a baby in a car seat, with your own personal roll cage. For normal adult drivers, the 3 point harness method is the only realistic method when you have no idea of size and shape of the people who will be driving a vehicle.

Finally...very few crashes result in fire, and not that many crashes are into water. I don't suggest NOT being able to self-extricate from a stuck belt, but it's not a scenario that is all that common. A Res-q-me tool or a lifehammer is a perfectly good tool - don't hang it from your rear-view mirror (I see that a LOT)if you intend to actually be able to find it after a wreck. It needs to be affixed somewhere in the vehicle.

They all do that, the question is when one is subjected to more than the design load would you prefer it to fail open or fail closed?

It is going to fail open when you exceed the tensile strength of the webbing, regardless.

So strike "design load", and insert "webbing tensile strength"

Now can I get a properly designed buckle?