[aprssig] Re: Tracker Smart Pathing: user types, alternatives

[apratt at bestbits.org]

One thing that makes it hard to talk about proportional pathing or
other ideas is that there are different categories of users, and they
want different things. So if one user judges a system good and another
doesn't, it could just mean they have different goals.

I think Bob's view of proportional pathing is based on a certain kind
of user: "I wish to be seen by other APRS users over a wide area, and
the receiver's level of interest in my position increases with their
proximity to me." Proportional pathing seems perfectly designed for
these users.

It's not as helpful for another kind of user: "I wish to be seen by a
particular other user over the air at particular place (e.g. SAR or an
event HQ), which will be an increasing number of hops away as I travel
farther afield." Once you get out of simplex range, every one of your
"direct" packets can be considered wasted. Once you're out of range of
the local digi, all of the one-hop packets are wasted too. Did
proportional pathing help you?

Here's another user (me) with different needs: "I wish to be seen on
findu as I travel widely in both urban and remote areas." There are
lots of reasons to want this; mine is domestic tranquility, so my wife
knows I'm still moving and haven't crashed my motorcycle. Proportional
pathing is not so good for me either.

In town, my direct packets have to get lucky to hit an IGate, but my
one-hop packets will have a very high probability. But in remote areas
there are two factors working against me. It's hard enough to reach a
digi in the mountains or deserts of California; the LAST thing I want
is for a successful "hit" to die before it reaches an IGate because
its hop count was too low. With proportional pathing, 88% of my
packets (seven-eighths) have a hop count of two or less. I don't like
those odds.

I know, I know, nobody's forcing me to use it. But I think a
meaningful alternative has been proposed: have digis rewrite hop
counts to the "right" value for the digi's local area.

Which routes can be rewritten? A route like WIDE2-2 is left alone; one
like AUTO,WIDE2-2 is a candidate for rewriting. This is keeps things
compatible, right?

What does this mean in areas where the infrastructure doesn't do this
for you? The WIDE2-2 part of the path will govern.

What hop count should such a digi write in? Ideally for my case, the
hop count would be "enough to (probably) reach an IGate." (No
guarantees.) That same number is probably a good value to use for
other likely "auto" users, but I don't understand their needs as well.

Who decides? I'd say: the same people who write web pages suggesting
what routing mobile users should use. The result is as if I were
constantly changing my route string to the recommended values as I
pass from one digi's footprint to another.

What about the problem of needing constant updates due to changing
(unreliable) IGates? Bah. This is a hop count, not a specific route
that needs constant attention. In terms of timeliness, a digi that's
updated once a year would be no worse than most advice found on web
pages.

If a digi can be programmed to TRAP high hop counts, then it should be
possible to bump UP the count of a packet if it's unsuitably small for
the area the digi is in. That's all I'm suggesting.

-- Allan Pratt, apratt at bestbits.org