<div dir="ltr">Open ended question:<div class="gmail_extra"><br><div>An email provider/directory will, at least, have:</div><div><br></div><div>1) A sender's IP address</div><div>2) Some sort of address for routing messages to their destination</div><div><br></div><div>Let's augment this with:</div><div><br></div><div>3) An authenticated message from the destination user specifying whether a given message is spam (and potentially, alerts for false positives if we deliver all messages but deliver spam messages to a spam folder)</div><div><br></div><div>Is it possible to build an accurate model of whether or not a given message will be spam based on these signals alone?</div><div><br></div>-- <br>Tony Arcieri<br>
</div></div>