Recently, with mucho fanfare, TLA launched a new service called inlinks – designed specifically to restore freedom to the link buying industry – freedom from the prying eyes of the grand overlord Google, freedom from… well actually just from Google.
I, being ardently against the domestic terrorism that fear, uncertainty and doubt marketing campaigns embody, would hail this as … whatever the modern equivalent of ‘groovy’ is - I’m too lazy to look it up in the urban dictionary.
Then I logged in, and my world crumbled. Wow that was way too dramatic. Okay - actually my muffin crumbled, as my jaw slackened, and bits fell in my coffee. That was the extent of the crumbling going on. It’s a butterface. She’s got a beautiful front-end, a beautiful back-end, she’s beautiful all around, everything but her face.
So I searched Inlinks for butter. I’ll spare you the details, but if you haven’t seen the system yet, herein lies the rub: they return not a URL (obviously, this is a closed inventory), but the PR, Alexa ranking and … wait for it … waaaaait for it … a contextual snippet of the text on the page containing your keyword.
The gears in my neocortext start-a-whurrin, and I realize, heeeeeey, who’s pretty good at associating snippets of text with URLs? Goooooooooogle. So I grab a snippet, “hack” some quotes around it, query it into the Godosphere, and kapow, I have a URL. Correlated to PR and Alexa ranking, I’m quite certain I’ve found where my link would live.
This is a very automatable process. Google is rather well known in some circles for their ability to automate processes.
Now, some of you are realizing, perhaps I’m being overdramatic with all my talk of domestic terror and muffin crumbs, and no, Google is likely never going to query the inlinks database for the phrase ‘exoskeleton Christmas gifts’ or whatever it is you’re optimizing for this holiday season - But chances are they *will* query for ‘christmas gifts’, and the niche exoskeleton site you’d found will be exposed.
An yes, hopefully TLA would be able to know if their DB were being over-queried, or strangely queried, and lock it down, but regardless, if any one of us can track-back a snippet to the most likely URL, Google could find a way to do it en-masse.
Am I dissing inlinks? Hardly. I’m kept awake at night for fear of Google penetrating my feeble mind with satellite based brain-rays. I simply think inlinks could be refined a little. They could either not provide a snippet for searches, or they could present the snippet as an image, and the inventory would instantly become much more closed, as it is intended to be.
Posted by Naoise Osborne said on 25.11.08 @ 12:51 pm :
Okay so since I wrote this early yesterday, they’ve significantly shortened the length of the snippet displayed in Inlinks. The column is still wide, but the characters returned are fewer.
Good work Inlinks - that’s effectively closed the loophole I’d say. It does make the snippet rather pointless, as you can no longer get a sense of the context of the phrase in question (which certainly did have some value), but it’s less exploitable.
Posted by Guillaume said on 26.11.08 @ 11:39 am :
I think that providing the snippet within an image would help more than shortening it!
Posted by Oggy said on 26.11.08 @ 11:59 am :
Well adding an image for each random snippet containing a word would not be easy, how many images would they need to produce for each article? If I type in “potato” would they create the snippet image on the fly?
Unless they showed the whole article on an image, and the database would query the text version and bring up the whole article in image form, which compromises the DB to prying eyes, I don’t see how they could scale creating “image snippets” for all their inventory.
Posted by Guillaume said on 26.11.08 @ 1:01 pm :
I am sure they could do some sort of automated snippet fetched from “8 words before the keyword + keyword + 8 words after the keyword” and generate it on the fly.
I don’t think it would be a major issue. People would still be able to find the queries manually by typing it in Google, but Google would have to hire zillions of monkeys to do so!
Posted by Gab Goldenberg said on 26.11.08 @ 5:07 pm :
Zillions of monkeys … you mean, quality raters :D? Or just the folks who respond to Matt Cutts’ call to report paid links?
A fictitious conversation…
SEO: Hmmm, I think our competitor is buying links.
Boss: How do you know?
SEO: Well, he’s got these awesome quality links in these aged blogposts that show good signals in Google…
Boss: Well, maybe he’s just well networked. Or doing linkbait. How can you prove it’s not legit?
SEO: Well, let’s check InLinks. I only need to show that 3-4 of these are paid to make the rest lose their value by reporting it :D.
On a related note, I’m making a pretty crazy offer you folks might care to take me up on - check the URL I’ve linked my name to.
Posted by Eyal said on 27.11.08 @ 9:06 am :
Good catch for that loophole.
Just providing my tech abilities to the discussion:
-It is easy to produce an image from text. There are some tools that already do that. Problem is - Google is probably able to hack that using OCR technology.
-Another solution would be to protect the data with human validation CAPTCHA(one of those scrappy texts that even humans hardly read). Google uses that to protect thier data in some of their tools. The better CAPTCHAs (like Google’s and Yahoo’s) are tough to break.
-Another thought I have regarding this is that if I were Google I would give lesser points for links that appear some time after the original text. So, if there was “miami hotel” within the content which became a link to some site, this is more probable of being bought link.