It's the best Sushi in West London, but when the owner entered into a deal with some other stores to create a common brand to help increase awareness and sales someone must have reported it as closed.
It's the same staff, same menu, same decor, same everything... except the name.
As I use Google Places a fair bit I felt that others might not go because of the listing, so I've been trying to fix their listing without a great deal of luck so far.
At least this conversation reminded me to click "Not true" yet again. Perhaps today a Googler will read this and 're-open' them.
I've been through the postcard validation, and it's pretty painless. I'd prefer some notification options that includes the ability to receive an email before my location is marked as closed. A thresholding algorithm could be used to detect abuse. If a location is repeatedly marked as closed, but the owner of the listing verifies that they're still open every time, a listing could be flagged for further inspection. Subsequent requests to close could be prompted for more information, which Google staff could then use to take action.
All of this requires more attention from Google, and I'd imagine that the Maps business already runs on pretty thin margins. Google's approach to most of these problems is to use an "85% solution" (I mean that 80% figuratively, not literally). That is, they choose a solution that works for "most" situations, but falls apart when abuse is present, but they don't really care, because those edge cases don't affect their bottom line.
For claimed businesses they can more easily contact you in case of a closed notification (they have your email address even) and should not take someone's word unless they can't get ahold of you.
For example talking about the operational details on how exactly things are verified would just make gaming the system easier. They probably don't know the exact magnitude of the problem, it would take valuable engineering time to come up with even a rough approximation, and it should be obvious that it's better to not comment than to just guess. You probably couldn't give a coherent explanation of why the "not true" doesn't stick without going into very tedious details of the data pipeline and serving infrastructure.