Machine Learning - Randomness in Search Engine Ranking Algorithms
-
I believe, sometimes you may 'deserve' the first position, but get #3. And of course, sometimes you may deserve #3 and instead be #1. All due to a 'randomness factor' in search engine algorithms.
I've been holding this hypothesis for quite sometime. Really, it started when I learned about SEOmoz using machine learning to better investigate SEO best practices.
I suddenly found myself wanting to learn more about machine learning, and the implications of using it for SEO. I'm by no means able to utilize machine learning myself, but I it appears unsupervised learning would have a real chance of cracking search engine algorithms. Hey, even Stuxnet was cracked!
Surely Google/ Bing would know (and account for) this, right? We can agree they'd obviously prefer a highly skilled mathematician not be able to crack the code.
Therefore, I'm led to believe that search engines use some sort of randomness in their rankings. Maybe not much. Perhaps not all the time. But if a random percentage of search results, had a random variable of sorts included in their calculations... wouldn't that be enough to prevent the vast majority of cracking attempts?
Thoughts, opinions, criticism? Thanks.
-
Hey Ryan, thanks for the ideas. Respectfully, I disagree however.
With something as simple as 'rand(1,4)' in php you can generate random #s. There's no telling what Google could do with their algorithm(s). Okay, so it's (relatively) easy to do, but why would they do it? After-all, they have heavily invested in complicated models that do the ranking for them...
Well, they've heavily invested in complicated algorithms, and if cracked, bump- there goes there flagship product. I'm sure SEOmoz aren't the only people attempting to glean insight using machine learning. And perhaps, others don't like to publish their findings and tip anyone off.
Of course, I'm not an A.I. scientist, so perhaps I misunderstand how unsupervised machine learning works, but I'm under the impression that it very well can be used to logically explain all of those factors. I'm sure there's tons and tons of work that would go into it, but if you had the resources (and some do), wouldn't you do it?
This is why I wanted to post this question though- to get feedback and other opinions, so thanks Ryan!
-
Hi Donnie.
I enjoy your analysis but disagree with your conclusion. I believe search engines are working with an unimaginably vast amount of data. They evaluate hundreds of factors then assign different weights to each. Also consider the source data is constantly changing, and the best anyone can have is a "view" of the data at a given point in time.
There is absolutely no need for any random factor. In my naivety, I believe every last ranking result can be logically explained. We simply do not have proper access to the data, metrics and weighting involved in those calculations. Our educated guesses seem to do a pretty darn good job.
-
Darn, I can't delete this comment
Browse Questions
Explore more categories
-
Moz Tools
Chat with the community about the Moz tools.
-
SEO Tactics
Discuss the SEO process with fellow marketers
-
Community
Discuss industry events, jobs, and news!
-
Digital Marketing
Chat about tactics outside of SEO
-
Research & Trends
Dive into research and trends in the search industry.
-
Support
Connect on product support and feature requests.
Related Questions
-
Is The Moz Community Rankings broken?
Not as active as I would have liked recently but still visit and login, answering a question now and again, yet I can see my profile is still listed as unranked. until recently I was somewhere around 45-49 and have been for many years - the criteria appears to be what it has for the last ten years or so. I see people with lower moz points outranking members with far more points. I can even see one with 0 points now appearing in the top 50 so my question is - is the moz community rankings broken or have I missed something? I know once before I pointed it out and it had to be fixed...input from all welcome 🙂
SEO Learn Center | | Matt-Williamson2 -
How to make google crawl our repository to make our site rank but make sure users dont go to our repository ?
We have a website that has links to documents related to various sectors. But the challenge is we do not have the documents on the website itself and they are linked to our document repository that has been blocked to google. We have put nofollow and noindex to the repository. Since Google can not read those documents, it has resulted in an impact in our SEO ranking. What would be the best way to make Google crawl the PDF documents in the repository at the same time make it invisible the "repo" not appear in the search engines. Would dofollow and noindex sequence work ?
SEO Learn Center | | PracticalActionDigital0 -
Can anyone share an example of a search strategy they've created?
New year, new plans. I'm currently working on outlining a search strategy for my company's marketing team. I'm primarily in charge of SEO but I'm working closely with our SEM and Content Marketing teams to build out a more holistic, integrated set of goals and initiatives for us all to work with. Does anyone have any good examples of a search strategy they've done in the past // a template they've worked with? I have a pretty good idea of what I want to include but would love to see how others have approached this + get some inspiration.
SEO Learn Center | | Jessie-beck1 -
Is anyone out there experienced in creating a blog template on Expression Engine?
I have a client that currently has their site on Expression Engine (EE). I have almost no familiarity with that, and am curious to know how user-friendly it is, and what is involved, for example, in creating a template to migrate a blog from WordPress to EE. I am getting more details, but in the meantime, wanted to find an "expert." Anyone out there know this CMS?
SEO Learn Center | | DianeDP0 -
What keywords will this page possibly rank for ... tool?
Hi all, I'm looking for a tool that will look at a web page that you submit to it and tell you what keywords it thinks the page is targetted towards and thus hopefully rank for. Does anyone have any suggestions please? Hopefully this type of tool will help us understand why page A is being ranked for keyword X when page B is actually the page that should be being ranked for that keyword. The only tool I have really found so far is http://www.ranks.nl/tools/spider.html I dont want to tell the tool what keywords I am looking at, as with Moz On page grader. Background reason (and possibly some one maybe able to shed some light on this also). We list a group of products... Nextbase Click 9 Lite, Nextbase Click 9 Lite Duo and Nextbase Click 9 Lite Duo Deluxe. All 3 are similar but very different products and each has its own page on our site. For the keyphrase "Nextbase click 9 lite" currently in Google we are listed for the pages containing the "Nextbase Click 9 Lite Duo" and "Nextbase Click 9 Lite Duo Deluxe"... but not the actual specific product page that targets the "Nextbase Click 9 Lite" product. Thus my reason for wanting to try such a tool that may help me understand why this is. Thanks for any hints, tips or pointers.
SEO Learn Center | | jasef0 -
Does canonical links rank in Google?
Our company has many pages that use the canonical tag. Will these pages rank in the search engine or will it pass the strength on to the original page?
SEO Learn Center | | WebRiverGroup0 -
Does Server Speed Effect SEO Rankings?
I was told by a website developer that Google ranks sites higher if they are hosted on faster servers. Is this true? For example my site is currently limited because it is a Quick Shopping Cart site hosted by Godaddy. He said that if I hosted my research chemical site with a different company with faster servers I could get better ranking immediately. I linked to my site. I think it loads ok. However Godaddy probably isn't the fastest service around.
SEO Learn Center | | chronicle0