Robert_Charlton - 1:22 am on Jan 14, 2013 (gmt 0)
Why do Google (at least their avatars tho are active on the forums) say that broken links, 404s don't affect rankings and crawling?
I wouldn't assume that this is exactly what Google is saying. If you're somehow creating the broken links and other 404s on your site, Google isn't going to like that because you're creating a situation that that wastes Google resources, and may ultimately reduce the number of useful pages on your site that can get spidered.
Analogous but not exactly the same, soft 404s (error pages which should be returning 404s but are in fact returning 200s), can hurt you. That's potentially an infinite number of urls returning "200 OK" responses. I think a large number of canonical errors (returning 200s) can also eventually hurt you.
What Google is saying that you shouldn't worry about are 404s that are not related to you... eg, from fragmentary urls that Google sees on other domains and is crawling. Google reports those in WMT, as "errors", so you know about them and can see if there's anything broken on your site.
For the above reasons, you should be returning 404s (or 410s) when called for... and if something in your site setup is somehow generating a great many 404s, you need to look at that. That doesn't necessarily generalize, though, to 404s from all sources hurting you, or even to 404s from a lot of urls you might naturally drop (as with classified ads) hurting you. (I think you can figure that if Google was crawling the pages in the first place, it will crawl the 404ed urls as well).
If you are able to implement 410s rather than 404s, return 410s for urls that you know are going to stay out of the index. Take a look at this discussion...
Best way to tell Googlebot a page doesn't exist anymore