Has anyone come up with a quick method for detecting if a domain
name is parked, but is not being used except displaying ads?
I'm hoping there is other method besides chasing a list of
constantly changing IP addresses being used by the parking
advertising companies.
a message of 6 lines which said:
Has anyone come up with a quick method for detecting if a domain
name is parked, but is not being used except displaying ads?
I don't think it is possible: "being parked" cannot be defined in an
algorithmic way. My own domain sources.org does not even have a Web
site (and I swear it is not parked).
Let's try:
* Bayesian filtering on the content of the Web page, after suitable
training?
* Number of different pages on the site (if n == 1 then the domain is
parked)?
* (Based on the analysis of many sites, not just one) Content of the
page "almost" identical to the content of many other pages? (Caveat:
the Apache default installation page...)
Stephane Bortzmeyer wrote:
a message of 6 lines which said:
Has anyone come up with a quick method for detecting if a domain
name is parked, but is not being used except displaying ads?I don't think it is possible: "being parked" cannot be defined in an
algorithmic way. My own domain sources.org does not even have a Web
site (and I swear it is not parked).Let's try:
* Bayesian filtering on the content of the Web page, after suitable
training?* Number of different pages on the site (if n == 1 then the domain is
parked)?* (Based on the analysis of many sites, not just one) Content of the
page "almost" identical to the content of many other pages? (Caveat:
the Apache default installation page...)
Dont forget there are mail only domains. I used to have one. Now it is
used to forward html somehow to my real homepage.
; <<>> DiG 9.1.3 <<>> -t any peter-dambier.de @212.227.123.12
;; global options: printcmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 28472
;; flags: qr aa rd; QUERY: 1, ANSWER: 6, AUTHORITY: 0, ADDITIONAL: 0
;; QUESTION SECTION:
;peter-dambier.de. IN ANY
;; ANSWER SECTION:
peter-dambier.de. 86400 IN SOA ns15.schlund.de. hostmaster.schlund.de. 2005050401 28800 7200 604800 86400
peter-dambier.de. 86400 IN NS ns15.schlund.de.
peter-dambier.de. 86400 IN NS ns16.schlund.de.
peter-dambier.de. 86400 IN MX 10 mx0.gmx.de.
peter-dambier.de. 86400 IN MX 10 mx0.gmx.net.
peter-dambier.de. 10800 IN A 82.165.62.90
;; Query time: 63 msec
;; SERVER: 212.227.123.12#53(212.227.123.12)
;; WHEN: Tue Aug 1 22:18:51 2006
;; MSG SIZE rcvd: 217
<HT ML><HE AD>
<TI TLE>Peter und Karin Dambier</TI TLE>
</HE AD>
<FR AMESET ROWS="100%,*" BORDER="0" FR AMEBORDER="0">
<FR AME SRC="http://www.peter-dambier.gmxhome.de/" SCROLLING="AUTO" NAME="bannerframe" NORESIZE>
</FR AMESET>
<NOF RAMES>
Peter und Karin Dambier
<P>
<DI V AL IGN="CENTER"><A HR EF="http://www.peter-dambier.gmxhome.de/">http://peter-dambier.de/</A></D IV>
</NOF RAMES>
</HT ML>
I have a large list of parked domains how would you like to query it and why do you want to?
-rick
Sean Donelan wrote:
* Sean Donelan:
Has anyone come up with a quick method for detecting if a domain
name is parked, but is not being used except displaying ads?
AFAICT, the main challenge is to define what "parked" means in the
context of your application.
Has anyone come up with a quick method for detecting if a domain
name is parked, but is not being used except displaying ads?AFAICT, the main challenge is to define what "parked" means in the
context of your application.
look at the time left on the parking meter
It seemed quite obvious to me: he's talking about domain squatting.
"Parking" is just a euphemism.
Parked:
A domain hosted by a middle-man for the sole purpose of generating
revenue from pay-per-click advertising. Characterized by having no
content of value.
This definition *might* work for NANOG, but my parking friends would disagree with the above.
Florian Weimer wrote:
In article <20060802200448.GA44921@icarus.home.lan>, Jeremy Chadwick <nanog@jdc.parodius.com> writes
AFAICT, the main challenge is to define what "parked" means in the
context of your application.It seemed quite obvious to me: he's talking about domain squatting.
"Parking" is just a euphemism.
I have domains (and over time most of mine have once been in this condition) which I've registered (for me, and no speculation involved) but not yet got around to publishing a website for.
I also have several where the published (and useful) website is just one page (harking back to a previous suggestion for a test of parking).
None of my sites has ever had any advertising (nor are likely to).
this needs to be "no original content of value"
BTW - for those who are still wondering about the question of detecting
this in semi-automated way, I recommend looking at what nameservers are used as way to determine if it is likely to be parked domain. Not perfect but you'll find large number of such domains and if that does not do it
then looking at common ip addresses of where the domain (www) is pointed
to will help determining this even more.
There seems to be DNSBL's for every other thing, I was expecting to find
one for parked domain names or the server IP addresses used.
This was for personal interest, rather than a commercial opportunity. I'm
a lousy typist and its unlikely change. But I can write computer
applications. I'd rather get a message my application can process
rather than relying on a human.
My preference is "legitimate" domain parking firms included a
standardized piece of meta-data my application could detect and use
as "this domain doesn't really exist." Sorta of a variant of the
web robots.txt file, but I prefer it to be application independent,
instead of assuming everything is HTTP Port 80. Perhaps start with a
standard record associated with the parked domain, i.e.
_notexist.example.com.
For less legitimate domain parking (i.e. typo-squatters), its a different
problem.
i know this will come as a shock, but there ar eother uses for domain
names than web sites
The trouble I see with this, is that legitimate web hosts commonly run several domains off one server, distinguishing by host headers.
So assuming that because 10 domains point at the same IP they must be parked, is a bad assumption.
william(at)elan.net wrote:
The trouble I see with this, is that legitimate web hosts commonly run several domains off one server, distinguishing by host headers.
So assuming that because 10 domains point at the same IP they must be parked, is a bad assumption.
It is. You need to have pre-determined (manually done) of what servers have these "parked" pages. Then if you see other domains pointing to
the same server, its almost always another "parked domain".
But as Randy points, just because its website is "parked" does not
mean domain is not being used in some other way. But in my opinion
this still qualifies domain as "parked" because common use of
"parked domain" term has to do with content of its website and does
not imply that domain is or is not being used in some unique way for
email or some other traffic.
Randy Bush wrote:
i know this will come as a shock, but there ar eother uses for domain
names than web sites
Surely you jest! Surely a domain with no listener on port 80 or 25 is not a legitimate domain.
Sean Donelan wrote:
Has anyone come up with a quick method for detecting if a domain
name is parked, but is not being used except displaying ads?AFAICT, the main challenge is to define what "parked" means in the
context of your application.There seems to be DNSBL's for every other thing, I was expecting to find
one for parked domain names or the server IP addresses used.This was for personal interest, rather than a commercial opportunity. I'm
a lousy typist and its unlikely change. But I can write computer
applications. I'd rather get a message my application can process
rather than relying on a human.My preference is "legitimate" domain parking firms included a
standardized piece of meta-data my application could detect and use
as "this domain doesn't really exist." Sorta of a variant of the
web robots.txt file, but I prefer it to be application independent,
instead of assuming everything is HTTP Port 80. Perhaps start with a
standard record associated with the parked domain, i.e.
_notexist.example.com.For less legitimate domain parking (i.e. typo-squatters), its a different
problem.
How about creating a database domain(domain_owner,domain_name)
and then querying by domain_owner. If the guy has more than 100 he looks
like a squatter and can me manually looked at.
e.g.
6.ag. 86400 IN NS ns1.sedoparking.com.
6.ag. 86400 IN NS ns2.sedoparking.com.
auktion.ag. 86400 IN NS ns1.sedoparking.com.
auktion.ag. 86400 IN NS ns2.sedoparking.com.
bilder.ag. 86400 IN NS ns1.sedoparking.com.
bilder.ag. 86400 IN NS ns2.sedoparking.com.
...
tvshop.ag. 86400 IN NS ns1.sedoparking.com.
tvshop.ag. 86400 IN NS ns2.sedoparking.com.
videothek.ag. 86400 IN NS ns1.sedoparking.com.
videothek.ag. 86400 IN NS ns2.sedoparking.com.
webhosting.ag. 86400 IN NS ns1.sedoparking.com.
webhosting.ag. 86400 IN NS ns2.sedoparking.com.
grep | wc says he has 51 lines. I guess it is 26 domains. The name suggests they are parked.
01.ag. 86400 IN NS ns19.schlund.de.
01.ag. 86400 IN NS ns20.schlund.de.
0800fitness.ag. 86400 IN NS ns11.schlund.de.
0800fitness.ag. 86400 IN NS ns12.schlund.de.
1-and-1.ag. 86400 IN NS ns3.schlund.de.
1-and-1.ag. 86400 IN NS ns4.schlund.de.
...
zusatzverdienst.ag. 86400 IN NS ns7.schlund.de.
zusatzverdienst.ag. 86400 IN NS ns8.schlund.de.
zweitmarkt.ag. 86400 IN NS ns25.schlund.de.
zweitmarkt.ag. 86400 IN NS ns26.schlund.de.
zypern.ag. 86400 IN NS ns21.schlund.de.
zypern.ag. 86400 IN NS ns22.schlund.de.
grep | wc says 3226 lines. But they are a famous german hoster. I dont think
they are squatting.
Just for curiousity AG is the german equivalent of PLC or SA in french.
I thought the namesevers would do. Maybe the whois gives more help.
Cheers
Peter and Karin
I have over 100 domains on my personal web server. _NONE_ of them are parked, although not all have web pages (and of the ones that do, none have ads).
The personal name server I run (in a group of several name servers we run collectively) has approximately 1000 domains on it. I can't guarantee that there is not a single parked domain, but the overwhelming majority are not parked.
I doubt we're "famous". How are you going to be able to tell they aren't parked? Pull up the web page on a few domains to see what they look like? Check all 1000 manually? Half?
Patrick W. Gilmore wrote:
I doubt we're "famous". How are you going to be able to tell they aren't parked? Pull up the web page on a few domains to see what they look like? Check all 1000 manually? Half?
Whose business is it. Who cares?
Although the original poster did not state a reason for why they wanted to detect such a domain - others have since suggested that the web site content on such a “parked domain” is of no (original) value since only ads run on such a site.
By that definition all billboards or stand alone advertising has no intrinsic content value. That complaint is justified only if you are lured into such a site under false pretense - such as by the site owner’s active efforts at search engine pollution - so the “offending” behaviour has to go beyond simply running ads on a “parked domain” to which you may not have been solicited.
Mistyping or typing in domain names and ending on such a site is a grey area - for example you dont blame the owner of a misdialed phone number for running any service they like on such a number just because it is two digits transposed from a “well known” or your otherwise intended phone number. That can go both ways - several cases of the wrong toll free number getting flooded with calls or the storied error from the 2004 US Presidential campaign when the Republicans sent the TV audience off to a Democratic leaning web site.
Yes, there are some speculators that are counting on user errors of omission or commission but an algorithmic divining of what the intent is is problematic.
Domain names are the “real estate” of the 21st century. You may wish to acquire a property for its “location”, rent it to someone else now, and only wish to use it for your own use in the future. You could just leave it unoccupied. This would only be considered a problem if you engaged in deceptive advertizing outside that property to lure someone in and tried to sell them something else.
That said, search engines do have their own heuristics on how to rank such pages “lower” in search results. Any articles that describe how Google’s page ranking works talks about ratio of native content to hyperlinked content, number of outbound links to inbound links etc, number of links to other pages on the same site (many “parked domains” are single page sites but the reverse is not always true)
Finally, if you have registered a domain lately - the web site associated with the domain is automatically associated with a “parked” page by most registrars (Network Solutions, Yahoo!, GoDaddy) immediately upon completion of registration and they run their own (revenue accruing to the registrar) ads on it till such time as you configure your own DNS servers and point it elsewhere. The maligned “middleman” comes into the picture later.
I am as frustrated as the next person when I end up on a site that lured me in with clever manipulation of keywords and search engine optimization - only to show me ads - but I would be loath to paint all “parked domains” with a broad brush.
I tried not to attribute malice on the part of domain parking operators.
I am looking for a way that you, or anyone else, could indicate a domain
should not be considered "in service" although the name is registered and
has an A record pointing to an active server so when I check that name
it doesn't require a human to interpret the results.
Most of the legit domain parking operators make it pretty obvious to
a human looking at the web page its not an active domain name , e.g. The
Future Home Of XYZ, Buy This Domain Now, etc. Unfortunately what may
be obvious to a human is sometimes difficult for a dumb computer. I
just want a way to make it equally obvious to a computer. As Randy points
out, there is more to the Net than the Web, so the better solution should
not depend on sending a query to port 80.