Forum Moderators: bakedjake
Does anyone know what sort of encoding this is, or have any ideas why the keyword was encoded? I'm unable to get a search on aol to show up with an encoded search term. I'm guessing it might be something obvious that I have just failed to understand ;)
Also a google search for enquery returns only 7 pages - mostly stats pages, and one page about "Runtime / Global / Dynamic Parameters" that is a bit too technical for me this afternoon :)
Anyway, try pasting the whole url into your browser, and it should tell you in the title bar what the search terms were.
I was interested in why they would bother to encode the search term at all. Also, if this is something aol are planning on using more often, it will make it pretty tricky to find out search terms from large numbers of referrers.
What's the benefit to AOL (or the searcher) in encoding the query?
it will make it pretty tricky to find out search terms from large numbers of referrers.
That's the only reason I can think of - privacy. Still, it doesn't make sense to me. The way the AOL cache/proxy works, it also makes it near impossible to track a user's click path through your site.
On another note, it could be a tracking technique though I do not understand it all very well.
Rob
Hopefully they'll 'see the light' and continue to give better referrers...
This involves two stages:
1. figure out the encryption algorhythm (of about 5 widely used)
2. Obtaining a key (the difficult part)
Does anybody feel up to it? The race is on, because i'm suspicious this is becoming a trend.
The way the AOL cache/proxy works, it also makes it near impossible to track a user's click path through your site.
this is possible, and it is available with at least one site I know, a visitor tracking service who can do this. Send me a sticky mail for info :)
Last thought: its probably not a privacy issue (AOL is not famous for protecting user privacy) Its probably just protecting their info - which has market value.
Its probably just protecting their info - which has market value.
You can use a cookie to uniquely identify the visitor which works pretty well and is much more reliable anyway in the long run IMO.
>>a script to query all the encoded refferers and find out what they were
That sounds like a good plan - i've seen a lot more of these referrers. Hope you'll let us know if you come up with anything (likewise wart888) :)
You can use a cookie to uniquely identify the visitor which works pretty well and is much more reliable anyway in the long run IMO.
Good point. I don't really like sites that set a cookie on every hit, so I only use them for shopping carts and such.
Does anybody feel up to it? The race is on, because i'm suspicious this is becoming a trend.
I'll give it a shot, but it'll be my first attempt at breaking a code. I did check to see if it's just the ascii characters encoded into hex. That's not it.
Adding or removing characters from the encoded string causes a 404 error. But, changing the value of a single encoded character causes a whole word in the unencoded string to be corrupted.
The beginning of the encoded string seems to correspond to the beginning of the unencoded string, and so on with the middle and end of the string. But, characters in the unencoded string must be hashed together in some way.
The encoding works in blocks of 8 characters of the original string at a time, turning that into 16 hex characters. This may be necessary so that they can cover the complete range of 16 bit Unicode characters.
http://aolsearch.aol.com/aol/ search?encquery=58310702BC93BBC6BC05B619A1E15263&invocationType =keyword
http://aolsearch.aol.com/aol/ search?encquery=3664A2C66E18E95E7AD5917A8D87D127&invocationType =keyword
they only use encoding for searches from people that use AOL as ISP
they only use encoding for searches from people that use AOL as ISP
I don't think so. Most of the aolsearch hits in my log are from AOL users, and most of them aren't encoded.
The encoding works in blocks of 8 characters of the original string at a time, turning that into 16 hex characters.
Yes, that's true.
How many more ads have they shown from webmasters checking keywords?
>>they only use encoding for searches from people that use AOL as ISPI don't think so. Most of the aolsearch hits in my log are from AOL users, and most of them aren't encoded.
Although they may not be using this for ALL aol user-searches, every example i've come across of an encoded keyword has been someone on AOL.
>>How many more ads have they shown from webmasters checking keywords?
Haha, I never thought of that, but I find it unlikely that this would be the main reason for encoding.
could be some are encoded and others not is due to the version of AOL being used. Maybe the newest version has this feature and others not?
are all the encoded terms 32 chars in length? this may be a tip you won't be able to decode, probably tied to an internal database of words/terms
could be some are encoded and others not is due to the version of AOL being used
are all the encoded terms 32 chars in length?
i feel somewhat bad about posting this, as i don't offer the full solution, but here goes anyway:
a) parse AOL-strings from log to data.txt
b) send strings off and recieve webpage
c) parse this string: <title>AOL Search: Results for "QUERY"</title>
d) dump QUERY to queries.txt
- you could do it with perl, php, asp, whatever... Anyway, be sure to put the script in the public domain and return an appropriate UA string to AOL, like "we want to see our users queries" ;) Personally, i don't have any visits from AOL.
/claus
probably tied to an internal database of words/terms
Thought of that, but no, because if you change one of the encoded characters, you get gibberish for the search string - the gibberish wouldn't be in the database.
No, I am seeing lots that are 48 chars in length as well, I have even seen one that was 80 long!
All multiples of 16, though.
you have the querystrings (refrr), and you can read the query off the AOL Search page.
True true, but it would take time and bandwidth to run. Might be the only way, though.
I saw a 16 character one too.
>>but it would take time and bandwidth to run
Exactly, and depending on what you are using to get stats might be pretty tricky to implement. Why do AOL always have to be different?
I saw a few more sites doing that.
Take the utf16 representation of the string, and juggle the bits any way you please... ;)
