Apart from the obvious technical point that referrer tracking isn't readily available, the idea itself felt a bit fishy to me back then, but I didn't quite pinpoint the problem. The problem in my mindframe being the following rather obvious fact: Referrers are bad.
How did referrers in http first come to attention? To me it was because they can be what makes web bugs work. Just place a tiny image from a central server on every page and that server can tell which page it is placed on. Nice for creepy service providers, a waste of bandwidth for everyone else.
Who benefits from the information gathered through referrers? Hardly anyone, I suppose. Within a web site, the information can perhaps be used to analyse the efficiency of the site's structure (how often is this really done? using referrer information?). But across sites? All that comes to my mind is the rarely seen feature to use referrer information for highlighting search terms when visitors come from a search engine. Otherwise, it seems to me like it mainly serves the vanity of web masters and bloggers.
In particular, allowing intra-site referrer information to be sent gives me as a surfer a tiny benefit at best and increased data throughput plus a loss of privacy at worst. So why should I send referrer information at all? When asked like this, many people will probably conclude that it's wiser not to, and next find out that they can't easily decide which information their mainstream web browser leaves throughout the internet without their knowledge.
That's a bad thing. And thus I am all in favour of sending no or misleading referrer information, as is done these days more frequently with the rise of RSS aggregators that lead to web pages being displayed in separate browsers with no referrer information transferred between the two. And I've even seen a case of referrer spam pointing to some porn site in our log files [you can detest those marketing-droids, but you've got to appreciate their efforts...].
What this adds up to for me is that sending referrer information doesn't give you any benefit while analysing referrer information only gives incomplete and potentially wrong data anyway. So just leave it. Send no or wrong referrers. You seem to be able to do that in Mozilla based browsers and iCab at least. Is there any trick known for Safari?
Off the top of my head, referrers have at least one more practical usage. If you host any images on your server, it’s possible for others to “deep link” to them — to create pages that include images hosted on your server. This can be really bad for someone who has metered bandwidth. At least twice in the past few months, I’ve had people on really heavily trafficked teenybopper web boards include images hosted on my server. In both cases, I was able to find this just by looking at my referrers, and block the hosts in question.
Again, there is no benefit for me as a surfer to send you referrer information. You put something publicly on the web, someone links to it… tough luck.
I like to consider that the beauty of the web.
I have seen that ‘randomly linked’ effect as well, with the sandal photo that I put up nine months ago to the day. It still gets about ten hits per day for no good reason. Not enough for me to worry about, certainly.
I get around 10,000 visitors a day.
When people deep link to me, I actually lose money - and deep linking to a flash animation can slow the server to a crawl.
So, dipsh*t, unless you can find another method to ensure people don’t steal direct links to my content, referrers are a necessity.
Received data seems to be invalid. The wanted file does probably not exist or the guys at last.fm changed something.