Hacker News new | ask | show | jobs
by toyg 4699 days ago
They could still do that in a secure way, client-side, if they wanted: when content is displayed in-browser, javascript could parse it, send home relevant words (on an encrypted channel), and receive relevant ads. The server would have to ensure that data is not saved, or it's anonymously aggregated right away -- you'll have to trust their word on that, but that'll always be the case.

Computationally expensive, maybe, but it's 2013 and browsers can take a bit of abuse. It wouldn't cover people using POP/IMAP, but Joe Average doesn't bother with that geekery anymore. Obviously it would take some time to implement, but it could be done.

2 comments

They'd also have to do spam filtering client side, parsing the MIME to extract inline images, attachments and so on, sanitize the HTML to protected against XSS attacks, process it for full-text search, filtering into labels, auto-forwarding and really absolutely everything that happens to an email. They all involve "reading" the email.
To be fair, that's all stuff that "real" email clients already do.
I think it's a little naive to assume Google just matches a few words in each email and serves ads based on that.
I'm not saying it'd be a 100% drop-in replacement for what they have now, but it'd be a good approximation, a starting point. I'm a leeching AdBlock user anyway, these days I don't even see what they try to show me...