To a first approximation, it seems that we've won the war on spam and lost the war on search engine optimization.
Most of what makes it to my inbox is at least somewhat relevant to me. In a lot of web searches, most of the first page is garbage, sites that are plagiarized from each other to trick people into watching ads.
@peterdrake What do you think the relative fractions of spam to real are in web pages and emails? Maybe the true-spam-detection rate is the same for both, but there are just more junk webpages? I agree that this still means we are not succeeding at filtering bad websites, but was curious what the reason is. It seems like we have more data for webpages in the form of the link graph, while we may have more personalization info for spam filtering for the user.
@kristinmbranson I don't have any data whatsoever, but my *sense* is that the vast majority of email is spam that is being filtered out.
Search results seem to have gotten noticeably worse in the last couple of years. Many of the hits are so obviously worthless that I can't imagine the link graph would rate them highly -- unless perhaps most pages are being built automatically, so there *are* tons of links to these paste-some-wikipedia-text-and-puts-ads-on-it pages.
@peterdrake I agree. I do use an adblocker, so I don't see Google's first page of ads. Interestingly, my eye is pretty good at ignoring the ads. E.g. I just looked back at the last search I did, which I thought it was an example of pretty direct results. Looking back on the page, there are 4 results before the relevant one that I didn't even remember. Seems like it should be a more solvable problem than it is. I wonder if Google etc. are incentivized to take us to ad pages.