Thoughts on GTFS Governance

[Brian Ferris]

I have a bigger question about governance of the GTFS spec.  I think Google has done a commendable job of encouraging and incorporating changes to the spec, but I wonder if a more official or open mechanism for incorporating changes and extensions might not further encourage adoption?  I'm not entirely sure what that mechanism should look like, as there are probably ten bad governance structures for every good one.  However, if the switch from Google to General is all about the Google in the name, the fact that the spec is hosted and ultimately still defined by Google will probably give similar pause.

So to get discussion started:

1) Are people happy with the current change mechanism?
2) Who should have final say on whether a new feature is added to the spec?
3) Is the current change mechanism visible enough to producers and consumers of GTFS?
4) Are the subscribers to the gtfs-changes list sufficiently representative of producers and consumers of GTFS data?
5) How much buy-in from a major GTFS stakeholder for a new feature is required for addition to the spec?
6) Should the change proposal be more formal than it is now?  Less formal?
7) Is Google as a gatekeeper to the spec a good thing?  Bad thing?  Just a thing?  If not Google, then who?
8) What role could GTFS play in defining additional data specs (real-time data, temporary service change data)?
9) Are there existing governance structures that could be looked to as a model for GTFS?

Like I said, I don't necessarily have answers to any or all of these questions.  Mostly, I'd just like to get a feel for what's working and what's not.

Thanks,
Brian

[Colin Bick]

Glad to see we've arrived at this discussion again. I have a little
input for some of the items:

1) Are people happy with the current change mechanism?

3) Is the current change mechanism visible enough to producers and
consumers of GTFS?

Mostly. Consider the "timepoint" field in the stop_times tables.
Despite implementations using this, and a push on this forum for
acceptance, this field never made it into the specification. I'm sure
there are good reasons for this -- for one, there is no model defining
the meaning of such a field -- but it hasn't been an open process.

2) Who should have final say on whether a new feature is added to the
spec?

A small collection of professional entities who cover the major uses
of the specification. For example: google, using it for trip planning;
a (tech savvy) transit agency, producing data in the spec; a consumer
of the data for uses other than trip planning.

8) What role could GTFS play in defining additional data specs (real-
time
data, temporary service change data)?

Oh the possibilities... I think this shows why a slow change process
is probably best, and why it would be good to have a producer of the
data involved in decisions.

[J. R. Westmoreland]

A list/group like we have now is a good idea and if we as a group adopt the
idea as a good one probably it should be made part of the spec.
See the IETF as an example.

I have ideas that would make my applications easier from a data use
specification but after talking it over with a few people here can
understand why it is the way it is. So, a group like we have here I feel
would be a good starting point as a governance group for the specification.

Just my couple of cents worth.

Best,
J. R.