Hacker News Reader

Comment by dreamux | Hacker News Reader

While I can appreciate (in fact, I probably don't) the technical difficulty of delivering a chat/video service to a userbase of 750M; these simply aren't compelling innovations for most (especially younger) people. Of particular note, these features aren't sticky (the usability of these features is common to _many_ other services) nor do they raise particularly high barriers to entry for competitors (Google has already nailed these services at large scales, integrating them into G+ is probably trivial). What FB (or Google for that matter) needs to nail is social search based on the graph of what your friends are doing/reading/buying... this is one area where FB has a distinct early advantage (they have more mature social graphs). I'm convinced social search will be the single feature that displaces or entrenches FB, not chat.

View on HN

Most people's social graphs are too small (130 people) to be useful. Well we will use an extended social-graph, you might say, but if you are extending the social-graph, why not include the entire internet? Why would my social-graph contain better (more accurate and relevant) information about doing a thing than doathing.com?

The social-graph is a reduced set of the internet. It is probably easier to search, but how will the results be better than Google?

There's a lot of room between the average user's social graph and the entire internet. I think if you got the best 20% of the internet compiled intelligently based on your social graph, you'd be in good shape.

>but how will the results be better than Google?

You have control of your own social graph, you don't have control over who puts a page up for Google to crawl. You're using the authority someone has as part of your network to filter what comes in to the network - search only on sites that friends-of-friends have recommended and you're unlikely to get spam or sites that are hard to navigate or that won't provide a good service. If I'm looking for new outdoor equipment I could look at the online stores that fellow scouts have used and recommended and so save time weeding out the better stores.

Just my immediate reaction to your question. There are clearly benefits in both types of search. Also this presupposes that FB aren't trying to inject advertised offerings in to your social-graph based search; probably a bad assumption as that there looks like gold to me.

I think you are mistaking how to integrate the social graph into search. You don't do it by exclusively searching within the social graph, you use the social graph to augment your traditional search. The best way for me to show you what I mean is for you to look at the attempts blekko is making. Create an account, and connect your facebook profile, wait 1 hour, and then do searches with "/likes" appended. You'll still get mostly normal results but the instances where a friend has liked a site will be featured more prominently. Facebook has not been able to "just make a search engine" with the social graph because making a traditional search engine is a non-trivial challenge.

Bing has a similar feature and one that might have more life given MSFT's relationship status with Facebook.

Hasn't google already taken a huge step towards this with their +1 initiative (which is obviously tied to G+)?

With Google's velocity, I think they're probably going to get to "social search" first (good luck to Blekko, but reliance on FB is not going to scale - especially when FB looks to monetize - social search is their meal ticket).

> this is one area where FB has a distinct early advantage (they have more mature social graphs).

Doesn't not having a search engine trump having a big social graph when it comes to search? Building search engines is not trivially easy. It took Microsoft 10 years to get it right.

this is exactly what www.blekko.com is trying to do.

I thought what blekko was trying to do was offload the work of building a search index onto its users.

Index technology is very well-known and pretty straightforward. You crawl the web. The quality of your search results are a function of your algorithm, and the depth/breadth of your index (created by this crawling). Blekko is attempting to outsource the algorithm part. Rather than play the cat and mouse game with spam sites, why not let power-users define what the best is, and then only return those results to others. Integrating the social graph is actually an incredibly powerful and compelling way to do search.

Basically returning results and ordering by the +1s your social graph has possibly already marked?

If I search for "2011 movies", that sounds great.

If I search for "digital camera", it might help a bit but if I were looking for the opinions of my photographer friends, I'd just ask them.

If I search for "us air strikes in libya" it might help but I would expect an algorithm to be able to find me some decent news articles without too much help from my friends. Besides, I don't want to live in my friends' 'filter bubble'.

If I search for "AT91SAM9G20 datasheet", I doubt anyone in my immediate social graph has even seen what I'm looking for, let alone +1'd it.

If I search for "AT91SAM9G20 datasheet", I doubt anyone in my immediate social graph has even seen what I'm looking for, let alone +1'd it.

That's true, but then how many competing documents are going to pop up? The social graph helps give precedence when their are many results.

The Libya example is interesting. If you want outside of your bubble there is no reason you couldn't reverse the order and try to see news that none of your friends have seen (or at least +1ed).

Reply on news.ycombinator.com