The "unintended side effect of filtering spam" explanation makes no sense.
How likely is it that the same company which builds bleeding-edge machine-learning systems to track and predict our behavior online, and which uses these AI predictions constantly to maximize their ad revenue, somehow cannot find a better way to filter out spam invites?
How likely is it that the same company that houses the likes of Hinton, Norvig and Kurzweil under the same roof can't find a better way?
Google is packed with experts at solving the "spam filtering" (i.e., pattern recognition) problem.
It appears this was done on purpose[1], driven by a corporate culture that no longer cares as much about openness. [Please read jholman's responses below. He's right, I went too far with this last sentence.]
--
[1] http://mail.jabber.org/pipermail/operators/2013-February/001...
--
Edits: added "it appears" at the end, to tone down the language. Also, reworded and added sentences to make my point clearer, and corrected text to refer to invites, not messages (thanks for pointing that out, mdc!) and point out that this was indeed done on purpose.
How likely is it that the same company which builds bleeding-edge machine-learning systems to track and predict our behavior online, and which uses these AI predictions constantly to maximize their ad revenue, somehow cannot find a better way to filter out spam invites?
How likely is it that the same company that houses the likes of Hinton, Norvig and Kurzweil under the same roof can't find a better way?
Google is packed with experts at solving the "spam filtering" (i.e., pattern recognition) problem.
It appears this was done on purpose[1], driven by a corporate culture that no longer cares as much about openness. [Please read jholman's responses below. He's right, I went too far with this last sentence.]
--
[1] http://mail.jabber.org/pipermail/operators/2013-February/001...
--
Edits: added "it appears" at the end, to tone down the language. Also, reworded and added sentences to make my point clearer, and corrected text to refer to invites, not messages (thanks for pointing that out, mdc!) and point out that this was indeed done on purpose.