[Peace-discuss] on counting traffic to test demonstration sites

[Stuart Levy]

Karen and Neil had offered to survey other street corners to see
whether, for example, University and State would be a better
demonstration site than Main & Neil.   Here's one measurement scheme --
this is what I did at University & Neil.

Things worth measuring:
   - how many cars pass in each direction (average per minute, say)?
   - how many cars stop in places where we can safely offer them a flyer?
   - how many flyer-reachable cars stop for long enough that we can reach them?

   - of the 8 corners at a typical intersection,
	which are suitable for flyering?
	* It might be possible to flyer on both sides of a one-way street.
	* If there's a turn lane, we probably *can't* safely cross it
	  (and the police will object if we try).
	  We can flyer to people who are stopped in the turn lane
	  itself, but only if they're waiting on a stoplight --
	  not if they're just waiting for a gap in the traffic.


Recipe:

With traffic in 3 or 4 directions, I had too few brain cells
to both count passing cars and also count stopped cars.

So, roughly every 10 minutes, I'd switch:

      - count all east/west passing cars until the light changed,
	  then write the number down,
      - count all north/south passing cars likewise until the
	  light changed.
	(Some cars turn, of course.   I didn't wait to see
	 whether they would, but just counted them according
	 to how they approached the intersection.)

	So you end up with a list of car-passing numbers
	in each direction for each light cycle.

     After about 10 minutes, note the time, and switch to
     counting stopped cars:

	- While the light's red in some direction,
	  see how many cars stop on each flyer-able
	  side of the street.  When the light turns,
	  note the count(s).

	  Repeat until about 10 minutes pass,
	  then note the time.  Then you get a list of
	  car-line queue lengths.

	  Ideally, we wouldn't count stopped cars at the
	  *end* of the light cycle, but maybe 20-30 seconds earlier,
	  so as to see how many stick around long enough to
	  hand a flyer to.  But I don't know how to do this.

At the end of this, you can divide the passing-car numbers
by time differences to get cars per minute, and average the
stopped queue lengths to guess at flyerable cars.

If you can, try to estimate how long the stopped cars stay stopped.
This seemed dramatically shorter, for southbound Neil traffic,
at University than at Main -- probably because University's stoplight
is timed for smooth flow coming from Main a block north,
while the next stoplight north of Main is many blocks away.
But I don't see an easy way to make a quantitative estimate.