Social search is resurfacing as a hot topic of late, due to how effective Twitter has become in helping you find information, and how it is close to how we source information in the offline world (via our network). Twitter is being differentiated by being called a "Help Engine".
I think it's getting us closer to the KM productivity (sense-making) aim that knowledge sharing and knowledge transfer has always aspired to, which is:
- finding the right information at the right time
- re-frame that information to be usable in your context and situation
- by connecting you to a social network of people you trust who will be willing to help out in a reciprocal relationship
(which also helps out in the re-contextualising process as you share a common wavelength or level understanding with people in your network) - learning organisation, information re-use, and corporate memory
From a particular perspective, the search experience is broken into three aspects:
- searching the web, searching within a website, and searching our network
- clarifying by reading and writing comments, and trackback/linkback blog posts
- asking a question within a website, within our network
Search for:
- facts/reference
- research a topic
- learning
- latest news about a topic
- looking for a particular thing
- you may not know if it exists or not
- you may have little information in hand to go by (exploratory search) - ask a question
- to find something
- about stuff you have found to get more context
Socialness to findability list
Google's PageRank is based on a referral model, so technically this is social search
- also comment/rank search results, also see Wikia
* based on whole web
Recommendations that are implicit based on your participation eg. Amazon recommendations
* based on content in a website
Google Blog Search (also Technorati, Backtype) is similar to PageRank
- but the point here is that the content is blogs which are a social ecosystem where you have distributed conversations (trackbacks/linkbacks), and leave comments
* based on whole web
Tag aggregators like Technorati Tags
- blog posts from the web filed by author generated tags
* based on whole web
Topic/blog aggregators
- a selection of curated sources
* based on another person's selection
How-to sites like Instructables
* based on content in a website
Topic sites like Squidoo, and topic wikis
* based on another person's selection
* based on content in a website
Review sites like Blippr
* based on your social network
* based on content in a website
Regular sites that have user reviews like TripAdvisor
* based on content in a website
Wikipedia is different to a regular topic site as the whole web can communally grow it
* based on the communal (people generated) web
* based on content in a website
Google Reader extends the concept of Google Blog Search in that I can create vertical people engines and subscribe to them.
- when I research a topic I search my Google Reader to see what the blogosphere has to say (often reading a blog post, links you to another, then another, and finally thanks to social linking you may just hop to the place you need...and lots of unexpected gems along the way...it's about the journey as well
- if I need more I can leave a comment on a blog post
- you can now converse with your friends
* based on your selection
* based on your social network
Bookmarks like delicious (or even YouTube, etc) are a social search as you are searching in the human indexed web
- you can even add people to your network and search in the bookmarks of your social network
* based on content in a website
* based on your social network
Blog networks like Vox and LiveJournal
- contribute, network and search within the website, and also within your network
* based on content in a website
* based on your social network
Lifestreams like Friendfeed consolidate all the profiles you have on the web, and allow you to network with others
- this social search is more than just delicious (bookmarks) or Google Reader (blog feeds) alone, this is searching all these profiles in the one go (it searches a lot more information)
- you can even comment on these items and discuss with your network (rather than just discussing directly with the author)
- you can even create Friendfeed rooms (which as similarities to a forum)
* based on content in a website
* based on your social network
Facebook is similar to Friendfeed but all your profiles are inhouse
- a big difference here is that in order to add people to your network they have to add you back (so this is closer to the friends idea, and is a big factor in building trust, reciprocal help, and having a similar level of understanding which helps with knowledge transfer)
- an advantage over a community of practice is you have a connection to weak ties
* based on your social network
Lijit is similar to Friendfeed, with the addition of searching your friends friends
* based on content in a website
* based on your social network, and extended social network
The ChaCha search model routes you to the closest expert to guide you in your search
- this is not quite the social search we have in mind, it's more like a reference librarian
* based on an authority/expert selection
Forums and Communities of Practice like Ning are conversational portals of information with people you often interact with and help out due to the reciprocal nature of a community
- there are strong ties here, so off-topic information will be hard to source from your community members
* based on your community
* based on content in a website
Micro-sharing sites like Twitter (Twitter Search)
- similar to blogging networks
- items in your stream are all at the same level it doesn't distinguish between a blog-type post, comment type-post, conversation-type post, question-type post, answer-type post...see my post on Twitter compared to other tools like IM and blogs
- your stream will also include posts that your friends are having with others
- close to real-time information
- you don't necessarily have to friend each other in order to follow a person's content, but like Facebook it does have an element of social bonding, trust and willingness to help out one another
- probe, clarify, discuss to learn and re-frame, and also a better chance of transfer if you are familiar with each others ways...this is contrary to unhelpful scenarios
- not just searching but you are daily learning (also advantage of weak ties)
* based on content in a website
* based on your social network, and extended social network
Mahalo Answers is a Q&A site
- there is more trust here when someone in your network offers an answer...also more reciprocation
* based on content in a website
* based on your social network
Yedda is a Q&A site
- there is more trust here when someone in your network offers an answer...also more reciprocation
- the difference here is that the question is also emailed to registered users who are the most likely to offer an answer (based on some sort of data mining)
- and if I emailed to help someone out, do I really want to spend time doing this for someone I don't know
- will I be burdened by my expertise
* based on content in a website
* based on data mining
* based on your social network
As you can see web 2.0 is the people generated web, and Google is no longer the first place we visit to search/find information. It's now Google's job to be aware of these sites, and lead people to them, as they are doing...you often now see Wikipedia and YouTube hits in Google results, and now Twitter hits.
Sure we can go to Google to get fast and ready results for quick facts
eg. today in a legal agreement I didn't know what the term quite enjoyment meant.
But if I want to probe deeper I would perhaps have to search blogs and leave comments or search my Twitter network (people I trust and share) and get a more timely result, from hopefully my weak ties.
The blog post, 6 Reasons Why Twitter is the Future of Search - Google Beware has some things to say:
"Twitter search is the ultimate social media platform and will enable people to get the opinions of others and add context to relevant information"
"Searchers don't just want facts. They want to learn more about the experiences of real people they can relate to."
"For example, rather than doing a search in Google for "best restaurants in new york" and getting a bunch of review sites, you can do a search on Twitter to see which restaurants people are talking about in New York. If you don't like the results, you can easily ask your network and get personalized answers in real time - which will then show up in future searches on the same topic.
Compare that to Google. They've been unsuccessful thus far in implementing social factors into the search results via Search Wiki. If you do a search in Google and can't find what you're looking for, what are you going to do? Probably ask around on Twitter."
The post also mentions your Twitter network as a solution to filter failure, close to real-time, localised around geographic location, information is representative of the masses, social network and trust "if you had a question about life in the NBA, would you rather ask Shaquille O'Neal on Twitter or type a question in Google?". I would say this is more likely effective if this person is in your network. People do have an altruistic nature in Twitter, but it's human nature that we will help out others that we know will help us out in the future...a lot of this has got to do with the limited time we have in a day...see the shadow of the future concept.
Help Engine
Danny Sullivan calls Twitter a Help Engine, which is obvious by the title of this blog post, How We Search With The Twitter "Help Engine". His post is not about the Twitter Search site, but how the Twitter site itself is used as a help engine, rather than in the past using something like Google as your first step. He also makes clear that using it as a help engine is a by product, as it's primary purpose is a micro-blogging network.
I know myself I'm lazy and just ask my Twitter network a question, but maybe it's not laziness, may it's effectiveness. And again I can clarify my answer.
Here's an example I asked the other day, I didn't even think to go to Google, I unconsciously went to Twitter, and got an answer 10 minutes later.
In Danny's survey the top answers in order were:
- Trust
- Expert answers
- Real-time response and news (hash tag conference tweets as it happens)
- Variety of opinion (of those you trust compared to Google results)
See the whole list below:
"* Fast answers, faster than searching and reading answers
* Easier to use when I'm mobile for answers than searching
* Too lazy to search
* Trust my friends and followers more than search results
* Want answer from particular person
* To get expert answers
* Because I couldn't find an answer on a search engine
* To get answers to "real time" issues (is Gmail working? is Time Warner Cable broadband broken again? Was that an earthquake?)
* Because I can follow up easily with further questions
* To get a variety of opinions rather than a specific "correct" answer
* For help finding something (article, news, web site) heard about but can't remember or locate"
Danny related Twitter to our information foraging before search engines came along:
"General search engines simply don't allow you to ask questions of friends en masse, something that was a top search habit until search engines came along. Twitter uniquely does allow this."
He also mentions that the effectiveness of Twitter as a help engine depends on your followers, so this makes a big difference in effort compared to Google. A comment from Danny's survey:
"As I have a small number of followers, I just don't have the base yet to expect to get a response to any question I might tweet. On the other hand, I have often done searches at search.twitter.com to gain insight into what the hive mind might be thinking on a particular topic or to get an answer that is timely that I would not be able to find on a search engine otherwise."
"I like using twitter to ask questions that involve personal opinion rather than straight facts. Often I can then follow-up with people as to why they say what they say, rather than the website author who may or may not be available for comment."
To all this I would reprise what Danny said in that the Help engine aspect is a by product...Twitter is a tool where we can socially bond, perpetually learn, express ourselves, feel connected and recognised.
In another post by Danny Sullivan, The Rise Of Help Engines: Twitter & Aardvark, he examines a study where we are returning to people we know and trust for opinions and context on subjective matters we are looking into. Here's an excerpt:
"...search engines can fail when it comes to subjective questions. What type of computer should you buy, a Mac or a Windows PC? A search engine can point you at resources such as computer reviews, but none of these resources will know the correct answer for your personal situation. That will be down to you.
It might be that you'll trust some of the resources you read. But often, you'll trust the opinions of those you know more. If a friend has a Mac, loves their Mac and encourages you to buy one after listing a few good reasons, that can shape your opinion.
If only there was a way to quickly ask all your friends for their advice and get answers back as quickly as doing a search on Google or another search engine. Then, perhaps, friends and family might trump search engines as an information resources.
Well, there is. That's the new revolution that's going on, a new type of search engine that effectively indexes the knowledge of those you know, so that you can query and get quick answers back from those people..."
Enterprise
This is highly relevant for the enterprise with tools like Socialtext signals, Socialcast, Yammer, etc...
Not sure why we need an ROI when all we are doing is using a tool that builds on the phone and email to help us do our work more effectively - learn, find people/information, conversations. We are all knowledge workers which means we are no longer of the methodology of "robots in a factory line process", we are moreso thinkers and have social interactions based on informal practice to get our work done...before a decision is made there are lots of converstions.
There are so many reports on people wasting time in looking for information (making sense of the workplace), and now we have tools that use people as a filter to better find information and people; and all the bonus learning, conversations, re-framing, adapting that happens along the way...federated search is good, but it won't do all that.
Why spend an hour looking for something, and to perhaps find that it doesn't even exist. A tool like Yammer can stream a question to the whole organisation (public timeline), and someone's network (if they follow you). It won't annoy them with an email, if they pay attention to the stream they may offer an answer...hopefully someone in your network is looking. If not, it was no harm trying, and it didn't cost any time or money to write a 140 character long question, and it didn't interrupt or bother people, it's up to them to look at the stream (which can be an accessible desktop app).
So now in addition to your search strategy you can also pop a question in Yammer "does anyone know if we have an official travel expense form...my federated searches are not finding it"
Hopefully within a couple of minutes you can stop searching as the Yammer help engine is using people to help you out.
Boyd's law describes this new paradigm (it's how we work offline anyway, but it's new to the online experience)
The whole reason for this post was to comment on a paper by the Aardvark team on social search, and it seems I have felt I needed to establish the landscape before reviewing their paper.
So both Yedda and Mahalo Answers seem to have most of the elements of a help engine (trust/reciprocation/context and clarity/high understanding of each other)...only Yedda layers something on top by also sending your question to the most likely person in the service based on data mining (the trust/reciprocation/understanding aspect is not as strong here).
But they do lack the close to real-time and mobile feel of the Twitter stream, and the fact that I can do many things on Twitter, one of them being asking a question. In the end no matter how good a service is, if it doesn't have enough users or if it's not where your network hangs out, then it's not gonna be much help...Twitter has the advantage here, as they have "the" audience.
From what I can gather Aardvark is similar to Yedda in the way that it finds the most suitable person for your question.
BUT, the big difference is when you ask a question it will be ported to a possible expert that is within your network, or your extended network (friends of friends), and it will also offer you a person who can connect with you at the moment in real-time (IM or the phone I suppose), otherwise in email or Twitter.
Danny Sullivan explains how it works:
- when you join you tag yourself with expert tags
- the person that invites you also tags you (reminds me of Lotus Connections Profiles or fringe) - set how often you want to be interrupted to answer questions (interrupted is not a good word, but anyway...)
- send a question by IM (or email which is less used, and SMS and a toolbar are in the works...but IM is better as it's chatty...even snail mail)
- await an answer or a conversation with a friend, or friend of a friend (it can base this on your facebook friends that are also registered at Aardvark...Twitter and LinkedIn coming soon)
Let's review how Aarvark achieves our sense-making KM aims.
- connecting you at the right time with some who may have the right information
- since you are connecting to a person you are able to re-contextualise and get lots of peripheral information which helps you relate their experience to your situation at hand
- this is even more effective when this person is someone you trust as they will take out the time to helps you (shadow of the future), and there is more chance their signal will be transferred to you as you understand their way and wavelength
Where it fails:
- the organisation by default or indirectly is happy because you are not duplicating, you are re-using information, and you are learning off one another and becoming more capable and adaptable (due to being able to tap into the organisation as a network vs a siloed hierarchy)
BUT, this is just on a demand basis, others are not getting a piggyback benefit by seeing your transactions in the open - The transaction is not a corporate memory as it will be happening in IM, or email...unless the receiver blogs about it
What's powerful about transactions in an open space like Twitter is that lots of people see it, and you aren't really interrupting them, they can choose to not help you, it's just another item in the stream
But at least an open stream allows people to learn daily, and build relationships, and allows others to eavesdrop and gift an answer from an unexpected person
Put simply it will miss out on answers by others that system didn't recognise
Why not have your question auto-tweeted as well, just in case?
Aha, VentureBeat alludes to a public view soon:
"Despite being quite powerful, the service still has its faults (after all, it is in beta testing). You can't search and view other users' questions/answers on Aardvark -but this is something that should be available when the site is public."
One clever thing it does, is that it may recognise your weak ties to help out with an answer, which is what sometimes happens on Facebook and Twitter, only Aardvark makes sure this is more than a potential by actually hooking you up.
It also gets around the need of having a massive follower count before it becomes useful, more from VentureBeat:
"Fred has more than 7,000 users on Twitter, I have barely 200 followers â€" but with Aardvark it not only evens the playing field, but opens the possibility of much, much more because of the network effect: The more users on Aardvark (there are currently more than 1,500 testing the beta), the more knowledge is available, and the faster the response.
Aardvark brings that power to the masses, and it leverages the collective intelligence (like Yahoo Answers) in real-time (like Twitter) without restricting you to just your followers."
I like this bit:
"You'll see more about the people you're interacting with, you'll see how you're connected socially. The routing algorithm will start favoring friends-of-friends, which is a very cool experience."
That's so true, your social graph is organic, sometimes your friends introduce you to a friend, and you end up becoming closer friends
Insight from their paper, What is Social Search?
We simply go to people to find answers, rather than a database, the better networked we are, the more we can tap into the right people.
"Most people rely on the human knowledge of those around them on a regular basis: when wandering over to a coworker, emailing a friend, or calling a family member, people are getting information that is personalized, timely, and trusted."
I really like this point; organisational know-how issues are not always about knowledge sharing per se, but more on understanding the current knowledge we have...this is the transfer part. Sometime we don't even realise the stuff we know ourselves, until it's triggered in conversation or seems to come out of us when faced in a new context. So there it is again, that word "conversation", the more we can converse the more the signal can be understood, and this has even more chance when your dialogue partner is someone you know well.
My KM 2.0 model also suggests that as long as we have participation and conversation, everything else follows. So the aim is not to manage knowledge, but to create conditions for conversation for knowledge to flow.
"Social Search extends this process in two crucial ways:
- By using automated indexing, Social Search products are able to determine exactly who in your extended network â€" among thousands of friends-of-friends and tens of thousands of peers with common affiliations â€" might have the knowledge you are looking for.
- By intermediating the process of asking for information, Social Search removes the social cost of requesting help from others. Concerns about interrupting or imposing on someone, or not having time for a long conversation, or spending social capital, etc., are all relieved by using a Social Search tool which initiates and manages the interaction."
This second point, above, is again related to the burden of expertise.
"Social Search is great for subjective questions, and questions where context is important to getting the information you want. Social Search complements Web Search, which is great for finding objective facts and public information.
Social Search is the right solution when what you want is a quick conversation with a real person who can interpret your question and understand what you are looking for. While Web Search can instantly provide you with millions of documents potentially related to your query, it cannot tell you which one is relevant to your personal needs in a specific context: just try searching for a hotel in London, or a
restaurant for a date, or great new music."
"...the vast majority of people do enjoy sharing their opinions and helping acquaintances in the real world. Most people have deep knowledge about a surprising variety of subjects. Finally, individuals generally are flattered to be asked for help; they enjoy having the chance to express their ideas; and they find it gratifying to be thanked for their assistance."
Nancy Dixon has a great post on this called Thankyou for sharing your knowledge.
"Social Search thus addresses the unmet needs of both consumers and producers in the information
marketplace:
- People with questions can access the knowledge in others' heads at the exact moment when
they need it. - People with answers can share the things they already know, thereby helping acquaintances
and making additional social connections."
"Social search is:
- Personally relevant: Social search is based on your social network...
- Contextually relevant
- Conversational and easy"
The Aardvark blog post, Social search is the new search, has more:
"Consider this: I have about 200 friends on Facebook, and they each have about 200 friends. Altogether I have over 10,000 friends and friends-of-friends in my extended network. These 10,000 people have a lot in common with me: many share my school and work affiliations and my cultural reference points. I'm interested in the choices they make and the experiences they have â€" they are usually more relevant to me than the opinions expressed by anonymous strangers on the web.
That's why Social Search is especially great for subjective questions, and questions where context is important to getting the information I want. We've noticed that Aardvark users find Social Search to be complementary to Web Search, which is still great for objective information that can be found on a specific web page."
Link to original post