Forum Moderators: open
Trusted TLDs
.EDU and .GOV - Trusted Top Level Domains
[webmasterworld.com...]
The primary document that I used for reference in the above topic is also going to be used for this discussion entitled...
THE - Trusted Hosting Environments
The specific patent I used for my supporting documentation was filed by Google on 2003 December 31. It is referred to as United States Patent Application 20050071741 and is titled...
Information Retrieval Based on Historical Data
[google.com...]
Within the above patent application, is this little tidbit of information which I've found similar references to in other technical documents related to search quality.
[0101] Also, or alternatively, the age, or other information, regarding a name server associated with a domain may be used to predict the legitimacy of the domain. A "good" name server may have a mix of different domains from different registrars and have a history of hosting those domains, while a "bad" name server might host mainly pornography or doorway domains, domains with commercial words (a common indicator of spam), or primarily bulk domains from a single registrar, or might be brand new.
Note the term "name server". Hosting environment is imperative in establishing a Trusted TLD. This applies to all TLDs and not just the naturally Trusted ones like .edu and .gov.
Personally, I feel that your Hosting Environment is an important part in the overall process of establishing yourself online. Take a look at those top positions in your industry and carefully review their hosting environment. There are a variety of tools you can use to determine what type of server the site is on, what type of OS, how many sites are hosted on that server, etc. Use them to your advantage when choosing a host.
Trusted Hosting Environments is a term that I coined after researching various aspects of what could potentially affect my client campaigns. I'm fortunate in that I've had a quality host from the time I started in 1995. I've been with three providers in that timeframe and 2 of the 3 were excellent. I now have my own servers and a team of administrators at my beck and call. I realized that if I was going to do this right, I needed to first establish a Trusted Hosting Environment.
Comments? Questions? Experiences? Anyone?
On a side note, hiding contact information in your Whois data may be a signal. I've never seen a need to do this and I personally think it can work against you if other signals are present.
[edited by: pageoneresults at 3:56 pm (utc) on April 25, 2006]
I am trying to balance out conflicting thoughts. Whilst in general I could easily give more "trust" to something in the .edu or .gov space, I have also been mulling over the fact that the most amount of logspam that I see is on .edu sites, and I wonder if that occurs simply because they are more lax in letting it happen than other types of sites are, or whether it is the spammers that specifically target those sorts of domains in order to falsely gain more "trust" than they might get from other types of domain. In any case, maybe those .edu domains are given too much "trust", or maybe their owners need to be better educated about how their space is rapidly being polluted and warned that maybe soon they will be given less "trust" by the algo. Whenever that happens, there will be a very big shift in search engine results arising from it.
I have recently notified several .edu sites about their logs being spammed, and their mailing list archives being polluted with v*agra link drops and suchlike, but they don't seem to be very aware of the ultimate consequences of not cleaing the mess up.
My emphasis...
On a side note, hiding contact information in your Whois data may be a signal. I've never seen a need to do this and I personally think it can work against you if other signals are present.
A lot of the spammier link networks I've traced ended up at feeder sites with anonymous registrations, very often with the same registrar... and I'm sure that Google has noticed this kind of thing too. I'm thinking it's likely that private registration raises some sort of a flag, but I doubt that it's actionable by itself. There can be good reasons to keep registration information private, and there's probably no way that Google can act on this one signal alone.
Reminds me a little about the questions regarding Google's noarchive tag and cloaking, and whether use of the tag would flag you as a cloaker. I think GoogleGuy reported that Google did take a look not long after the tag went live and noted at the time there was a huge correlation between use of the tag and cloaking. I'm not a cloaker, but I remember some follow up discussion about whether it might not be best, if you did cloak, simply to cloak without the noarchive tag... that your chances of being discovered would be less. I think it's clear now that sites aren't losing rank simply for using the tag... again, perhaps, "unless other signals are present." And to me cloaking without the tag seems foolhardy.
So, to look at private registration in an analogous way, how much should we hesitate to register domains privately? The idea of not being on a lot of lists has its appeal.
Regarding the original post, I'm noting that while the Google patent focuses on "name server," pageone raises questions about shared hosting, a topic that from time to time has gotten a lot of discussion. To take it a step further... beyond shared IPs, need we be concerned about unique IPs at hosts whose reputation we don't know, or for that matter, at popular hosts whose reputation we think we do know?
On the other hand, his post may have said all that anyone wants to discuss in public.
I had to wonder why this topic didn't get much attention. Then I thought, how many people would actually investigate this? How many would know how to?
A lot of the spammier link networks I've traced ended up at feeder sites with anonymous registrations, very often with the same registrar.
I do believe this is very easy to detect even for the simplest of search engines. ;)
I'm thinking it's likely that private registration raises some sort of a flag, but I doubt that it's actionable by itself.
Me too and yes, I would also doubt that it is actionable by itself. As you say...
There can be good reasons to keep registration information private, and there's probably no way that Google can act on this one signal alone.
So, to look at private registration in an analogous way, how much should we hesitate to register domains privately? The idea of not being on a lot of lists has its appeal.
I think the pros far outweigh the cons in this issue.
Regarding the original post, I'm noting that while the Google patent focuses on "name server," pageone raises questions about shared hosting, a topic that from time to time has gotten a lot of discussion.
I've not really done any testing in this area and wouldn't want to. I do know that some sites I've reviewed in the past where problems were present were being hosted on a shared IP with thousands of other sites, some in very bad neighborhoods.
To take it a step further... beyond shared IPs, need we be concerned about unique IPs at hosts whose reputation we don't know, or for that matter, at popular hosts whose reputation we think we do know?
I was. I am. I think IP Blocks can become poisoned. How to determine that is challenging to say the least. So, to avoid the issue altogether, I just got my own servers and my own IPs (unused). I've been on this mission to control every aspect of my client's website presence. I want to make sure there are no weak links in the system.
Surely the neighborhood you decide to move into is going to have some impact on your marketing efforts, wouldn't you think so? Guilt by association? ;)
2. the Top Level Domain:
* country code (ccTLD): some countries certainly appear to be treated better than others. A simple filter that can really hurt legitimate domains in a designated 'bad neighbourhood' ccTLD. Link value from some ccTLD's is definitely depressed and others appear to have a volume threshold to their value.
* generic (gTLD): covers the gamet:
- .mil and .edu are generally SEO gold (there is evidence certain directory names in the link negate/depress link value).
- .com, .net, and .org appear the most volatile with a possible white list (currently general link values appear to being depressed (by G) but those from 'certain brands' still work magic) and some evidence of 5+ PR linkage holding value.
- sponsored (sTLD), i.e. .museum, seem to be weighted well in certain circumstances.
- unsponsored (uTLD), i.e. .biz, .info, .name, seem neutral with some evidence that certain numbers cause link value depression.
Each TLD is easily filtered - expect dramatic change each time one is targeted whole and less dramatic change when another filter, i.e. directory name, is used to depress value usually given for TLD 'trust'.
3. the Registrar:
* not all Registrars meet my 'trust' level but I have not seen evidence that SEs care.
* Whois is a definite filter. Same registrants for linking sites appears to negate link value above a certain threshold. Invoking privacy frequently seems to be treated like cloaking. Links from such registrants sites appear of neutral or depressed (from expected) value.
* There is a viable but more expensive method: incorporate each site and list Whois info as the agent unique to that company. No evidence (yet) that SEs are linking company ownership with websites.
4. the Host:
* This is the 'trust' biggy. Certain hosting services are SERP suicide. Even with a 'trusted' host shared hosting raises the problem of 'just who are your servermates?'. You must do continual due diligence and frequently move from server neighbourhood to server neighbourhood or spend a bit and get a dedicated server/IP.
* IP address block is an obvious filter. If you find yourself in a suddenly 'untrusted' neighbourhood' moving is about all you can do in a hurry to regain/maintain SERP. If you suddenly get mucho links from such disaster areas block them as best you can to speed their demise. And try not to link into such areas; above certain thresholds this appears detremental.
5. the Domain Name Server:
* is your DNS data on an untrusted host?
Until your DNS host impliments DNSSEC (RFCs 4033, 4034, 4035 published March 2005) you must do your own due diligence.
* I have seen possible 'trust' values show in SERP changes shortly (within 30-days) following DNS location changes. Nothing definitive but sufficient to show in analysis. I was surprised to see possible SE 'trust' interest in this area.
The idea of 'trust' or weighted TLD filters is valid in my experience.
As these are simple to invoke and easy to change I expect continuing ruthless use by the SEs (especially Google). A very big hammer chasing a lot of small pinpricks.
On the other hand, his post may have said all that anyone wants to discuss in public.I had to wonder why this topic didn't get much attention. Then I thought, how many people would actually investigate this? How many would know how to?
For those of us who are interested in learning more about our servers and our neighbourhood (in the event of a site being hosted on a shared IP), I personally use Netcraft and DomainTools (formerly Whois.sc)
-----------------------------
Learning how to choose a hosting company is a skill in itself and with the plethora of companies around, here are a couple of tips that I thought I would share:
hope this helps
if anyone uses other tools or has anything to add, please feel free.
How much information do the likes of Google get from a whois? I ask as I manage several domains for other people under a single account. Does Google see that they are under one account with XYZ registrar?
If so, does anyone know of a registrar that has an option to automatically create a new user account (as far as whois is concerned) but still allows the user (me) to manage all the domains under a single login account?
They can see the registrar (who usually register 000.000's of domains), the registrant (who should be your customers), and the admin tech contact (who might be you or your customer).
I asked as sometimes I get sales (spam) emails for my main account with the registrar I use but which are targeted at a domain I manage for someone else and which has different whois info ("Let us get your XYZ site ranked number 1" type emails but sent to my ABC mail account which does not appear on the whois). So I concluded registrars or some people also get access to some form of account number.
