Omit InfoWindow in embedded Google Map

>> The locations for a particular latitude and longitude actually show up several hundred yards apart in GoogleMaps API access and  GoogleMaps).

Is this true for all addresses?

If you are feeding it the address it is probably caused by someone moving the marker.  See http://maps.google.com/support/bin/answer.py?answer=68474&query=move+marker&topic=&type=#modify

If you are feeding it the address it might be caused by different base data/algorithms for translating house number to distance along a street.

I wonder if that is caused by each system using different datum? For an example of the difference  see:  http://www.adaweb.net/devserv/survey/verifica/thirteen.htm   - doubt it, but compare the amount of shift caused by datum change. See lower corner of USGS topgraphic quad for the area.

I wonder if it is caused by different registration of the mapping to the aerial photography? I have noticed that streets that used to be missed aligned with the aerial photography are better aligned.

Not only that, but as they update their mapping the picture under the lat/log might change.

>> The locations for a particular latitude and longitude actually show up several hundred yards apart in GoogleMaps API access and  GoogleMaps).
>
> Is this true for all addresses?
>
> If you are feeding it the address it is probably caused by someone moving the marker.

I am specifying latitude and longitude, so it's neither the geocoding (mapping from address to lat/long) nor someone moving an address marker.  I really am trying to avoid as many possibly-error-inducing components as I can!

I suspect it's a registration issue, but also wonder about precision.  How precise is lat/long given by hours to six decimal places?  I can't imagine them choosing a standard too imprecise for the intended use, but do notice many of the differences in the thirteen.htm you reference are several decimal places down in the seconds.  Anyway, that's curiosity.  

Practically, mapping of lat/long differing by several hundred yards on the two systems puts my customer's site half way down the next block, which is not acceptable.  I will manually enter the correction for my customer's address so it's correct at Google Maps (when looked up by address), but that won't affect the embedded map marker location (which is specified by lat/long).

> Not only that, but as they update their mapping the picture under the lat/log might change.

What I meant by "yech; problematic as they improve their mapping" is that any compensation I make for the discrepency will backfire as their lat/long locating improves.  If there are registration problems, they're going to have this issue anyway (with all the address locations people have "corrected") as they improve registration, though they could update algorithmically their database of corrections to reflect the new locating with each improvement.

Sigh.  I am so far past the three hours I had allocated for this task :-).

Apparently a known bug, I mean feature  :-)       Google says

Why does the API Geocoder provide different locations than Google Maps?

The API geocoder and Google Maps geocoder rely on two different data sources. This is because the API tiles use a different data source than the main Google Maps, and we wanted to make sure the points returned from the geocoder would match up with the tiles displayed. The API geocoder occasionally gets updated with new data, so you can expect to see results changing or improving over time.
See http://code.google.com/support/bin/answer.py?answer=60738

I think what that means is that you need to Geocode (determine lat/long) in the same system (API or Maps) that you plan to use for display. i.e. you can not use maps.Google.com to determine lat/long to be used in the API,  

Sigh