Reverse Engineering Google's Algos

This is pretty amazing. I mean does everyone really think that google decided to reinvent all of their source code and algorithms in the last 5 years? or just tweak certain settings and add things. I mean you see updates like penguin and panda really could be just one small thing added to the search metric that disturbs the system as a whole which is huge in a database like google that contains billions of websites. Those changes seem so BIG to us and have huge hype around them all on SEO / Webmaster blogs that we forget that the actual change could just be a small change like dropping the weight of forum links to 3 instead of 4. I think your on to something here if you look back at where it assigns a metric value to certain LINKS, why would a company intranet really need "link weighting algos"? Because it is already too embedded into their source as a whole and they cannot get rid of it without messing up their whole search engine. Really worth digging into here. Do you think when they released this to corporate companies to purchase they though, "hey we better not, someone on BHW might have enough cash to buy this and tear apart our algo"?

I would take the metrics that seem to be in these files into huge account and just add a few things that are popular now, that weren't back in 2009 like Facebook and social links. well to googles eyes anyway. I like where this is going.
 
I personally wish to thank you. I have been looking for this for two years but did not know what it was called. Took less than ten minutes to locate once you named it. Downloading now. You do not know what you have!

Since I PM'd you, this is the difference between what I have and what you have:
gsa.PNG
This is from the dl.google link.

Incidently, here is the URL for the current Google Search Appliance API source code:
http://code.google.com/p/google-ent...ds/detail?name=gsa-admin-api-cs-1.0.2-src.zip

and the GSA API developers guide:
https://developers.google.com/search-appliance/documentation/68/gdata/acapi_java?hl=zh-CN
 
Last edited:
wow this is golden. Great thinking.
 
@Gophering - firstly I can't believe this thread has been hiding from me for 6 weeks, so whoever bumped it , thanks. And secondly I don't know where this will go either, but let me just say that even if it doesn't really go anywhere you (and the others helping) are in no way wasting your time. (+14 rep for all of you involved)
 
Last edited:
Great thread, you got the wheels in my brain turning. Following.
 
I subbed and +rep you. This thread is so interesting. I like the part on how Google determine links and pages, but I guess it will be close to impossible to find out?
 
Yeah or maybe the mods can setup an escrow account, where all BHW members with say...100 posts can contribute lets say $10 each.
If we just had 1000 members contribute $10 each there would be $10,000 in the pot to get the latest GSA, which would have the latest weighting of factors used to rank pages.



This is pretty amazing. I mean does everyone really think that google decided to reinvent all of their source code and algorithms in the last 5 years? or just tweak certain settings and add things. I mean you see updates like penguin and panda really could be just one small thing added to the search metric that disturbs the system as a whole which is huge in a database like google that contains billions of websites. Those changes seem so BIG to us and have huge hype around them all on SEO / Webmaster blogs that we forget that the actual change could just be a small change like dropping the weight of forum links to 3 instead of 4. I think your on to something here if you look back at where it assigns a metric value to certain LINKS, why would a company intranet really need "link weighting algos"? Because it is already too embedded into their source as a whole and they cannot get rid of it without messing up their whole search engine. Really worth digging into here. Do you think when they released this to corporate companies to purchase they though, "hey we better not, someone on BHW might have enough cash to buy this and tear apart our algo"?

I would take the metrics that seem to be in these files into huge account and just add a few things that are popular now, that weren't back in 2009 like Facebook and social links. well to googles eyes anyway. I like where this is going.
 
Very interesting thread, subscribed to see more about the algo :)
 
Hey, Gophering, it would be really cool if you posted the download link you found on baidu, so everybody can get the virtual machine image and play with Google Search Appliance. I, for one, would really appreciate it.

Anyway, thanks for the GREAT post.
 
People like OP are the future of SEO. Hopefully the reverse engineering will bring up some tasty little secrets..
Good luck for you and other participants with this project!
 
Yeah or maybe the mods can setup an escrow account, where all BHW members with say...100 posts can contribute lets say $10 each.
If we just had 1000 members contribute $10 each there would be $10,000 in the pot to get the latest GSA, which would have the latest weighting of factors used to rank pages.

Count me im on that

Sent from my SAMSUNG-SGH-I747 using Tapatalk 2
 
Hey everyone,

appreciate everybody's enthusiasm about the topic. I've (sadly) been quite busy at work for the last month or so, didn't really have any time to investigate further (or post any of my current findings). However, I have not completely forgotten about this topic and continue to dig whenever time allows. I hope to have some more time next week to post some of the juicier findings.
Stay tuned and thanks everyone.

Cheers
 
Back
Top