From G+ discussion where Chrome devs discussed this decision in May of this year.
The suggestion of rigging up monitoring to look for this was considered, but the case was seen as so obvious (user-state loss vs. a week or two to adapt to a changed shortcut) that it wasn't considered worth the effort.
IMO the data above is enough when coupled with anecdotal data. Given that it didn't seem worth investing the considerable engineering effort we would need to gather more data. Although there was disagreement on that point. As Peter said, this decision was only loosely based on metrics. Ultimately it was intuition based on a combination of the data mentioned above and anecdote.
The suggestion of rigging up monitoring to look for this was considered, but the case was seen as so obvious (user-state loss vs. a week or two to adapt to a changed shortcut) that it wasn't considered worth the effort.
See Ojan Vafai's comment(s) here:
https://plus.google.com/+KentonVarda/posts/F1dio2L9XtS
IMO the data above is enough when coupled with anecdotal data. Given that it didn't seem worth investing the considerable engineering effort we would need to gather more data. Although there was disagreement on that point. As Peter said, this decision was only loosely based on metrics. Ultimately it was intuition based on a combination of the data mentioned above and anecdote.