Most humans request the favicon.

Most robots don't.

That's as definitive as we can get.

have you examined the HTTP Response headers for the favicon.ico requests?
are you supplying the proper headers to support user agent caching of this resource?

> have you examined the HTTP Response headers for the favicon.ico requests?

Until I transition my site fully to https I am not able to log browser language or character set (because of IIS4). So that leaves me with user-agent and referrer (and requesting IP and method - ie get or head) as a way to judge if the requester is human or bot. Requests for favicon.ico give same logging info in that regard to the other files being requested during the hit.

> are you supplying the proper headers to support user agent caching of this resource?

I don't think I have any code in any html files that pertains to caching. I don't make use of htaccess files (I never see browsers request them - should they be?). I see both code 200 and 304 - which tells me that some browsers are just checking for the existence of files and not actually downloading them, probably because they've visited the site in the past.

do you ever see a 304 response to a request for favicon.ico?

...(because of IIS4)
Are you sure it is IIS4? if so, are you saying you are on NT box?

Oh never mind the above question, i just read [webmasterworld.com...]

After removing anything with "bot" in the user-agent (and removing all google and bing IP's) I get about 11,000 hits to favicon.ico (this is from jan-1 2015 to the present). Of those, practically all of them have http response code 200.

11 have code 206 (from 5 different IP's) and 80 have code 304.

While on the subject - I searched the logs for "Google Favicon". It turns up in a few different user-agents, such as this one:

Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/49.0.2623.75 Safari/537.36 Google Favicon

That agent first appears on March 11/2016 and continues right up to a few days ago. It grabs mostly favicon.ico and default.html and sometimes 1 or 2 other files.

Prior to March 11/2016 I see this:

Mozilla/5.0 (Windows NT 6.1; rv, 6.0) Gecko/20110814 Firefox/6.0 Google Favicon

which runs from Dec. 8 / 2015 to march 8 / 2016.

Prior to that, I see (just) this:

Google favicon

All told, about 1300 hits (from Jan 1 / 2015 to the present) from the Google Favicon. Do those hits mean anything? Is a Google Favicon hit indicative of something?

Is a Google Favicon hit indicative of something?

I don't think anyone has worked out a comprehensive list of everything Google's faviconbot is for. An obvious one is GSC: stop by and it will collect the favicons for all your sites. If you list websites in your G+ profile, those also trigger favicon requests whenever someone views your profile.

The one that I've yet to figure out is when the faviconbot visits my test site, which isn't indexed and of course doesn't have a GSC account because why would it. It isn't common, compared to “real” sites, but why is it happening at all? (Quick detour to raw logs reveals that they just started doing it in late September, and every single request comes in at HTTP, redirected to HTTPS, meaning that the front page is requested twice, the favicon only once. On average, every 2-3 days. Hmmm.)

Up until a few years ago, google's faviconbot sent no UA at all; you could only identify it by IP and behavior. And then, yup, FF/6. The current incarnation is at least something plausible.