Archive for the ‘Social Graph’ category

Scoble’s Social Graph Videos TRANSCRIBED

August 27, 2007

Mr. Scoble’s Trilogy on Social Graph Based Search is a fun watch. Recommended! (maybe helped to poke a hole in the ‘searchability’ theory though)

If you read the comments, there are LOTS of people weighing in that they don’t have the time to WATCH it. If you have an interest in SEO and SEM stuff… make time. I did. I think his last posit (the Yahoo! one) has a great deal of merit. You might not put 1 + 1 + 1 + 1 + 1 toghether to arrive at his same conclusions without the bennies of the previous 30 or so minutes. But, for those of you short on time, I’ve transcribed it.

Robert, if you’d prefer I take this down… I’ll happily comply. Likewise, if you want to run with it and flesh it out further for your own purposes… Go for it!

Social Graph Based Search

Good morning, I’ve been thinking about this since hearing Jason Calacanis at Gnomedex.
Theres a revolution coming in search, Social Graph Based Search. It will overtake Google, in four years.
You probably didn’t get here from search. Search doesn’t index video. and I’m intentionally anti-SEO’ing this post.

SEO resistant technologies…
1) Pages are studied by GYMA. Tags – Title, H1 & other headings, paragraphs
Results are built off of repetition (with gaming filters applied). Google’s version of this is PageRank. Paid links are indistinguishable from organic links. This came to a head at the SES conference.
2) Techmeme works differently than a search engine…
Started with 1,000 of the top tech bloggers with emphasis on linking behaviors. Built a fabric of websites (10-20K blogs).
Getting added to the fabric is based on a critical mass of buzz. The scooper gets top position on the meme. Hangtime on Techmeme is based on strength of signal.
SEO resistant because you’d have to be in the first 1K of tech blogs OR be trusted by same.
SEOs can’t break into the trust ring easily.
3) Mahalo even better… because 100 people trusted and paid to work with 1,000+ in community.
User generated content checked by the 100 and verify quality of content.
SEO resistance via the 100. Sanitized by nature of the business.
4) Facebook appears not to be a search engine. A social graph.
35M in the social graph. Matters because it’s all interlinked on a basis of trust.
Starting to sound similar to others.

Mahalo, Techmeme, Facebook covered in part I.
Some fit in search. Some don’t.
Mahalo’s absolutely going after search – Google specifically.
Just look at the HDTV results at Google. ALL of them mention HDTV in title and desciptor tags and probably have lots of inbound links.
What DON’T you see? Manufacturers.
But, what do YOU do first? What do you need to know? Know who the manufacturers are! The noise is wacky at Google.
Google has a profusion of services now.
Go to Mahalo and look for HDTV. You see lots of high quality, pertinent, useful links – top to bottom.
It’s still incomplete… But, high signal. No noise.
I’d have to refine my Google search to get close to this.
That’s TODAY. Give it four years and see where both of these guys are.
Google’s going to have to change FAST to understand social behaviors.
ORKUT’s not used in their own search and they own it.
The small guys come in and write their own rules and put a jujitsu on Google.

How would we build a new search engine ala these three?
H1, P and linking.
Bootstrap the old system. Right on top of it (“it” = Google).
Mahalo should add a Google undergirthing.
The humans are simply advanced searchers and will get better over time.
How about the trust level?
Jason -> The 100 -> fans out from there with ratings and high-grading.
Based on affinity, ratings, etc the power paging starts to take root.
From here Mahalo and Facebook can study one another and rank/weight things with some high confidence.
Facebook has figured out how to lock out the spammers.
Loss of credibiility is a simple matter of scratching out the entrants.
Techmeme has already built the fabric of trusted individuals. It’s pure and clean.
Four years because…
Mahalo, two years (Facebook and Techmeme) for mainstream to catch on. Mossberg/Markhoff will take a while to get to it.
Two years for people to move over AFTER the mainstream reviews weigh in. Classic adoption cycle.
Two years of obviously better search results are necessary to be the catalyst.
Inviting open debate via video or postings.

Extra bonus round
This stuff doesn’t scale well!
Google Reader example… at personal limits 5-800 feeds. Some of which are agreggates of other feeds.
Can read more than 1,000 items superficially. Not recombinating.
Mahalo’s probably at a certain threshold. Assuming a certain community size you yield a certain amount of pages created per day.
There’s some math checking needed here… Still. Some critical mass is being built in this model.
What if the algo allowed for everyone having their own little Mahalo?
If each of us made fabric based on our own N degrees of seperation… it COULD scale.
The 100 yields 30,000 items. Jason’s not going to be complacent with The 100. He’s going to grow from there.
“He’s a loud mouth!” -Scoble
Watch Yahoo! They have some weird mojo going on. They’ve invested in small things. They’re the wildcard.