Ernest Prabhakar wrote:

So, why not always choose the generic unit "census tracts"?   Does
anyone know exactly how those are defined?  They should be small enough
to be immune from gerrymandering, but easier to manage than block-level
data (which I don't think is always well-defined, anyway).

Matt replies:

Warning: I am a novice on this topic.  We want a unit that is resistant to being manipulated to favor the majority party incumbents.  My understanding is that census tracts are subject to modification every ten years.  If municipalities are more difficult to modify and mostly remain the same since they were created then that gives an advantage in this regard to municipalities.  If the process of census tract drawing can be kept isolated from political control then census tracts may be the better choice.  However, the only objective consideration in drawing census tracks that I am aware of (in the U.S.) is that they contain 5-10,000 persons, which leaves census tract drawing vulnerable to political manipulation.  

I was hoping that my questions would solicit some insight into whether municipalities can be too large for voting districting and if so, whether a fair procedure to split them can be divised.  Maybe a reliable deterministic procedure for splitting an existing geographic unit exists or can be devised, in which case this isn't a major disadvantage for using municipalities.  BTW, I found a glossary of usefull terms on http://www.westmiller.com/fairvote2k/in_gloss.htm.  

Ernest Prabhakar wrote:

Could you elaborate on why you don't think road traffic a good measure?

Matt replies:

Yes.  I have in mind long distance commuter, vacation and truck traffic which positvely correlates to road traffic without positvely correleating with neighboring community relationships.  Also, poorer or rural census tracts may have less traffic making them more prone to be divided.  Also, mass transportation would need to be monitored for "fairness".  Measuring traffic throughput accurately on all roads at census tract borders would seem to be an expensive and difficult undertaking and would still fail to take into account the critical issue of where the traffic starts and ends.  Also, bandwith alone as an alternative ignores the difference between expressways and small roads.  Some roads may have many exits inside a census tract others may have none.  Using Miles per hour as a compromise measure fails to distinguish between roads with many traffic lights or stop signs and those with none.  I don't think roads can be utilized as an unbiased or accurate measure of community relatedeness.

As for manipulable problem, road construction routing decisions appear to me to be partly a technical and partly a political issue.  You cite the democratic nature of the political process to defend it as unbiased.  However, democratic political processes tend to favor the "haves" versus "have nots".  Who has the most political clout will vary based on which party is in control since each party has different constituencies concentrated in different census tracts.  The technical considerations, such as flat versus mountain, grassland versus forest, swampy versus rocky, private versus public ownership, heavily built versus vacant, etc.  do not make the road construction manipulatable but they do also contribute to making road construction biased as a measure of community.