<div dir="ltr">I was thinking of the Seti@home app that used to be popular back in the 90's. Before that I always had to set the time on all my devices. In my experience my cell phone has updated time change info on something other than the tz database. I don't know how Verizon did it (unless it's through complex shape files), but when driving by Vidal, it was on Arizona time; 6 miles north it switched to Pacific time. I know because I let the tz database know about this strange little town which Verizon somehow knew about and kept time with (probably only a population of about 12).<div><br></div><div>Perhaps an iterative process like this could help improve accuracy when users change the time standard when it is wrong sending that info back to a server that the tz database reads?</div></div><div class="gmail_extra"><br><div class="gmail_quote">On Thu, Oct 16, 2014 at 4:25 AM, Jonathan M Davis <span dir="ltr"><<a href="mailto:jmdavisProg@gmx.com" target="_blank">jmdavisProg@gmx.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">On Thursday, October 16, 2014 03:49:56 Zoidsoft wrote:<br>
> This might seem like a stupid question, but now that most of the world is<br>
> connected by computers, smart phones, etc... why don't we develop an app<br>
> that collects time settings and standards along with GPS for every device<br>
> world wide as to what is showing on the clock? Such an app could then<br>
> automatically generate zone rules along with location data for each of<br>
> those rules. It would be a complex exercise in set theory getting the<br>
> program to recognize "groups" of similar data but I think it could be done.<br>
<br>
</span>The first question that comes to mind is how on earth each of those computers<br>
would get the time in the first place. Its the TZ database that this list is<br>
for that's used to give most of them the information that they need to give<br>
the time in a local time zone rather than UTC. All that stuff like NTP does is<br>
give the current time in UTC, so without something like the TZ database, all<br>
any of those devices would know about would be UTC.<br>
<br>
Also, time zone rules change over time, and we often need to know what they<br>
are ahead of time, so trying to figure out what they are via AI would not only<br>
be a mess, but it wouldn't even have the information that we need when we need<br>
it. If all you cared about was generating a database of what a bunch of<br>
devices _thought_ that the time was in a given time zone at different points<br>
in time was, that would be one thing, but we care about having the correct<br>
rules over pretty much as large a time span as possible - both in the future<br>
and in the past. And what you're suggesting wouldn't work for that at all. AI<br>
can be great for heuristics, but it's horrible if you need to know the exact<br>
anwser to something - like what the time zone rules are _exactly_ for a given<br>
time zone at any given point in time.<br>
<br>
But really, what you're suggesting pretty much comes down to generating<br>
information from a group of devices instead of looking it up in exactly the<br>
same place that they look it up, because the only way that they know anything<br>
about the local time zone that they're in is because of the TZ database. They<br>
don't magically learn it just because a government announces it. They learn it<br>
because their TZ database files get updated with it.<br>
<span class="HOEnZb"><font color="#888888"><br>
- Jonathan M Davis<br>
<br>
</font></span></blockquote></div><br></div>