they've said that they will start devaluing them in 2007, but I personally have a hard time believing they will know the difference on every site. Often more than not, on a site, it's completely undetectable.
See excerpt from Matt Cutts (the voice of Google): "Google does consider buying text links for PageRank purposes to be outside our quality guidelines" and that he "wouldn't be surprised if search engines began to take stronger action against link buying in the near future". He also revealed further information about what can happen to web sites that sell links: "Reputable sites that sell links won't have their search engine rankings or PageRank penalized–a search for [daily cal] would still return dailycal.org. However, link-selling sites can lose their ability to give reputation".
Then all paid directories will die . All paid blogs will live. There will appear two well-known paid blogs, one is aliveblog,the other is avivablog.
Not necessarily. If your directory is charging for a link placement, it may die. But if you are charging a "review fee" (yahoo does it), it will not as google will not have any valid reason.
Paid directories won't die, but many will become less profitable. I really wonder how many new paid directories pop up everyday. I have not used any paid directories, because there are plenty of other free opportunities, if you can be a little patient. I can only imagine a world without adsense, paid links, and Google's duplicate content filter actually worked. It would probably lead to more profits for me and my affiliates, but adsense and paid links will always be around. The good news is that Google might get their filter right one of these days.
Show me a way to detect (for certain) a paid link. There isn't one. I could show you 100 pages with "links". I bet you couldn't get more than 50% right, and you'd probably do far worse than you think. Phynder said it all..they've said it in 2004, 2005, 2006. You can't program the impossible.
If you get thousands of links to your site in a short time, Google might think you've bought links and hence de-value them. Also, if you ONLY get links from high PR sites in a short period of time, Google will again feel that you've purchased links from high PR sites. So, when you get links to your site, make sure you get links from both high and low PR pages / sites.
Wow! I would love to see that. Saying you can detect paid links is one thing - but VERY easily detect paid links? Fantabulous! Tell us all about it. Or better yet, prove to us that you can do it.
First collect all attributes of the backlink ,then construct a Multi-attribute Evaluation Model using SVM to determine which link is paid.
I don't think google can detect if a site's links are really paid. They would have to change their algo and put use a "Nostradamus" bot if they want to detect if the links of all the websites in the internet have paid links or not.
Its quite plainly impossible to do this (without a crystal ball). They can look for patterns etc to indicate paid links, which is one reason why I don't think going through a broker is a good idea - plus I have generally found that all you get is some listing in the footer of a non relevant site/page. However, if I go to jo blogs site and say I will pay you for a link on page X and give him some money through paypal, then how on earth can be detected? - Your telling me google keeps a track on my paypal account? To be honest paid linking has been going on since year dot and it covers almost every form of internet advertising - if you put a banner on a site its still a link.
i have a feelings that google is just talking about yahoo directory. they can easily de-value the yahoo links, but would have a very very almost hard time de-valuing all paid links. just my guess.
You must be new to the data mining world! If you were not, then you would realize that determining which attributes, then collecting the data on said attributes is a hefty amount of work. The easy (and fun) part is running the data through your SVM|ANN|GP|DT|NLR algorithm. Care to give your reasons for using a kernel based method for this task?