Advice for "DU this poll" threads.

[FatDave]

Any time you're on a website, which I'll call Site A, and a page on Site A has a link to another site, which I'll call Site B, and you click on it, scripts on Site B can tell that you came there from a link on Site A. Web browsers sent that information in the HTTP request.

Now, if I was running a poll on a website, I might detect when a whole bunch of hits were coming from the same site and filter them out, weight them differently, or something else. I'm not saying anybody is doing this, but they might be.

So, I would suggest rather than just clicking the link to the poll, if you right click it you should get an option like "copy link location" (in Firefox) or "copy shortcut" (in IE). Pick this option, and the URL from the link will be copied to your clipboard. Go to your address bar, clear what's in it, and press ctrl-v and then enter. Now you're at the poll page and they have no way of knowing where you came from.

Am I too paranoid? Maybe. But you never know.

[evlbstrd]

1. I have a better idea.

Go to a porn site first and jump from there.

[csziggy]

3. Even better - go to a freeper site and then jump

[evlbstrd]

4. I'd enjoy the porn site more.

But point taken.

[R. P. McMurphy]

6. Porn site/Freeper site...

What's the difference? They're both obscene!!! :evilgrin:

[FatDave]

8. It only matters if you follow a link.

If you enter the address into the address bar, nothing about the previous page goes over.

[Just-plain-Kathy]

2. Thanks Dave, now have to bookmark this post. :)

[dbonds]

5. Well, that would take an IP cross reference with the http log for them to do that.

For the initial page load they do get the referrer as part of the http request, but then to answer the poll question is almost always a form submit which is another http request that would be have the site the survey is on as its referrer. So the only way they could weight DU referred links different than others is if they went in and parsed the IP of the form submit against the http log that would contain the initial IP with the referrer - and that would be too CPU expensive for most sites to do.

[FatDave]

7. Not necessarily.

The main page could pass the referer along in a hidden for field. The main page could determine if the refering site was blacklisted and pass along a flag as a hidden form field or even as part of the url in the form action attribute. Or even have the form processed by a different script entirely. Chances are the site has some kind of session handling, if for no other reason than to make sure you don;t answer more than once. Flag it as a blacklisted url in the session, in the cookie, whatever.

With any kind of poll running on a website, there's a 90% chance you've already got the overhead of a database back-end, so you could easily handle all the blacklisting invisibly on the back end. Pop the ip, flag linked to the session, whatever, into the database, have the poll count script look it up. No need to churn through logs whatsoever.

If a client asked me to build a poll with the ability to filter out certain referring sites, I'd have half-a-dozen options available in a matter of minutes.

And these days, there's at best a 50/50 chance it's going to go through a regular form submit, what with javascript validation, AJAX and all.

[frustrated_lefty]

9. Better yet, ignore them.

They accomplish nothing beyond self-congratulation.

[FatDave]

10. You really think so?

You don't think there's anything to be said for our opinions being publicly viewed by a wide audience? Making our numbers known is meaningless to you? :shrug: