Ooh, I had this same confusion last year! "Proxied by" is all about approximation.
Y isn’t *exactly* X, but it’s close enough to use in place of X when X is messy or unobservable.
For example, in machine learning, you might proxy "user satisfaction" with "time spent on page." Not perfect, but useful!
Y isn’t *exactly* X, but it’s close enough to use in place of X when X is messy or unobservable.
For example, in machine learning, you might proxy "user satisfaction" with "time spent on page." Not perfect, but useful!
