[tz] Comments and mapping of tz zones to the real world

[Tobias Conradi]

On Sun, May 6, 2012 at 8:49 AM, Kevin Kenny <kkenny2 at nycap.rr.com> wrote:
> On 05/06/2012 02:03 AM, Tobias Conradi wrote:
>
>> Can you tell a way of how a user determines the correct zone for a
>> random location in Indiana, selecting between
>>
>> America/Indiana/Indianapolis
>> America/Indiana/Vincennes
>> America/Indiana/Winamac
>> America/Indiana/Marengo
>> America/Indiana/Petersburg
>> America/Indiana/Vevay
>> America/Kentucky/Louisville
>> America/New York
>>
>> without reading the comments?
>
>
> Fortunately, nearly everyone in Indiana is aware of the bizarre
> timezone patchwork.
What do you mean by "bizarre timezone patchwork."? Do you mean the
existence of eleven zones for Indiana?

Adding to the above eight, these three:
America/Indiana/Knox
America/Indiana/Tell_City
America/Chicago

I would assume many people wouldn't know that there are eleven tz
zones for locations in Indiana.

>
> The simple rule, "if you're in Eastern time with DST, use New_York,
> if you're in Eastern time without DST, use Indianapolis; if you're
> in Central time, use Chicago" is correct for 99% of them, because
> the breakaway counties are farmland with few inhabitants.

Using http://en.wikipedia.org/wiki/Time_in_Indiana#tz_database and
adding up Wikipedia 2010 population data for counties having a tz zone
that is not one of the three you singled out, namely Chicago,
New_York, Indianapolis, gives:

Pulaski	13,402
Pike	12,845
Daviess	31,648
Dubois	41,889
Knox	38,44
Martin	10,334
Crawford	10,713
Clark	110,232
Floyd	74,578
Harrison	39,364
Switzerland	10,613
Starke	23,363
Perry	19,338
SUM	436,759

Indiana population is given with 6,483,802 for 2010. That is the three
zones you singled out contain 93.3% of the Indiana population at most,
thus cannot be correct for 99%.

>  Out
> in the hinterland, it becomes "if you are in one of the half-dozen
> counties that have their own zones (named like the counties), use
> that."
I see no single zone that is named like a county.

The above county list contains two half dozens and one county. I don't
know what you mean by "the half-dozen counties that have their own
zones".

> If you were to pick a random location on a map in Indiana and ask
> what time is observed there, I'd be hard pressed to answer.
> But
> in that particular case, better mapping might not even help;
I was asking for how to determine the tz zone without using the data
in the comments, i.e. without any mapping for individual counties at
all. Because on  Sat, 5 May 2012 04:11:55 +0200 Robert Elz claims:

" erroneous comments are not a bug, as they don't affect the results"

> a great
> many inhabitants and businesses set their clocks to the large cities
> of Chicago or Indianapolis rather than whatever time their county
> legislatures have decided to observe that year.
Any example for this? It could be added to the tz database and improve
the mapping.

If only a minority in a given county does not observe what the law
mandates and is in the tz db comments, then having the comments
improves the results on average, thus the assumption "better mapping
[compared to no mapping] might not help" is wrong. If a majority in a
given country does not observe what the law mandates, I would ask on
the tz mailing list, whether the information for that area should be
changed.

-- 
Tobias Conradi
Rheinsberger Str. 18
10115 Berlin
Germany

http://tobiasconradi.com/