Hacker News Reader

I have a bet with random_walker that a distributed social network will gain significant traction by 2014.

http://twitter.com/#!/flipzagging/status/16766878672228352

Diaspora is a sad story: a team with great intentions but little systems experience, becoming the standard-bearer for distributed social networking. They gained attention from some random factors (frustration with Facebook at a peak, NYtimes story, Kickstarter).

I think the odds of there being a single winner-take-all provider of social networking services (SNS), forever and ever, are slight. The idea that kids are going to stick with the SNS their mom uses to play Farmville seems ridiculous to me. So we should see some amount of fragmentation, eventually.

It also seems absurd to me that Facebook will remain agile enough to shut out competitors while also supporting 500M+ users. At some point, a situation should arise where competitor SNS services will find it to their advantage to be interoperable. That could end in an oligopoly, but that still counts as a more distributed social network.

Larry Page often likes to say "don't bet against the internet". And I am old enough to remember when people thought that Microsoft's Project Blackbird was going to kill the web. I am confident that SNS, like almost all internet services, will be commodified at some point, and probably sooner than we expect.

Caveat: this hasn't quite worked out for instant messaging, so I could be wrong.

Those bet terms seem really odd. Your bet is that within 3 years, there will be a distributed social network with at least 1M users? Am I reading that correctly?

Google Plus, which people aren't even really sure is a success yet, had 10 million users in less than a month. Is there some reason that you chose the number 1M as significant traction? If anything, those terms suggest to me that you're not really convinced it's going to happen.

Google Plus is not distributed, and they gained most of their users rolling over from gmail.

I believe you misunderstood my point.

1. People are uncertain whether Google Plus, a centralized social network, will gain significant enough traction to become a force in social networks.

2. Google Plus, in 2011, has had extended periods (weeks, months) where it gained 1M users every 3 days (the number of users in the bet). Thus, orders of magnitude more than 1M users is not sufficient to necessarily be a substantial social network.

3. In 2014, presumably there will be even more users on the Internet, and it will be possible to gain traction even faster than in 2011. Further, a social network might need to be even larger to be a substantial force in 2014.

4. Therefore, why would one set the threshold for "significant" traction for distributed social networks in 2014 at one network with 1M users, when that number may be orders of magnitude too small to be relevant in 2011?

(This ignores the question, of course, of what counts as distributed and/or what counts as a social network.)

Obviously, Google Plus is not distributed, or else one of the people would have already won the bet, since Google Plus already has millions of users!

I am confused, however, by your comment about users rolling over from Gmail. First, do you have evidence for this claim? (If anything, it seemed to be false in the case of Buzz.) Second, how is it relevant?

Google Plus is a new service that goes with your Google account, just like Gmail, Documents, Maps, etc. They don't count as new users if they already have a Google account.

He's not saying Google Plus fulfills the bet, but rather that the user base it has amassed is an example of why winning the bet would be meaningless, or at the very least an unnaturally easy target.

I chose those terms to win easily. ;)

buddycloud already claims over 500K users and seems to have significant momentum, some developer discipline, and is using well-known standards, even if XMPP isn't everyone's favorite standard.

And as others mention, you might want to count identi.ca too.

If you think my bet is too paltry perhaps we can raise the stakes.

However, note I said greater than 1M monthly actives, which is a more serious measure of traction than just user accounts.

I think that fragmentation (esp. by interest, and even more so by profession) will make SNS much more profitable in the long run.

You may be close to winning already. Identica was at half a million (may not be active) users two months ago. Adding in StatusNet, TWiT Army, and all the other nodes it could be there.

http://identi.ca/conversation/75407544

I'd bet against you, but I don't think your definition of "gain significant traction" is reasonable.

It's pretty easy for a network to get 1 million sign-ups, and then it's pretty easy to get 1 million people to pipe their Twitter status's into it. See Identica etc.

That doesn't mean is has significant adoption as a social network - it's really just a Twitter client.

I agree that there will be many social networks, but I don't think there is much reason for them to interoperate.

Take a listen to Eric Ries talk about his experience building an interoperable IM client[1].

He built the whole thing, then showed it to heavy IM users and they were (paraphrasing) "err.. why would I want that? I already run 6 IM clients, running another isn't a problem. And I don't want the same people to see me on each one"

We already have an interoperable, distributed social network for messaging purposes. It's called email. I doubt another one is needed. (OTOH, I did build a Pubsubhub powered demo of one back in the FriendFeed days. The technology was cool...)

[1] http://ecorner.stanford.edu/authorMaterialInfo.html?mid=2273

Trillian http://www.trillian.im/ have been selling their cross-platform and interoperable IM client for over a decade.

If people didn't want Eric's interoperable IM client, it wasn't for lack of a market.

A niche market != significant adoption.

I know lots of people using Linux on the desktop too. It's a market, but isn't really significant in comparison to its competition.

Trillian is either a good or bad example, depending on your perspective. I don't want to have N IM accounts; I want to have one account/identity that can reach everyone. That has not happened with IM and I am skeptical that it will happen with social networking.

You guys should put your bet on www.longbet.org!

Caveat: this hasn't quite worked out for instant messaging, so I could be wrong.

Aren't Google Talk and Facebook Chat both based on Jabber/XMPP, though?

I wasn’t really aware that computer science went into research directions like that. It seems either like classic social science (in this case communication studies) territory or like business economics territory.

What approach are you taking when looking at budding and failed social networks? Did you do a quantitative study or were you looking for patterns in a more qualitative fashion?

It's a purely qualitative study.

You're right, it's a bit outside the normal ambit of computer science. I'm interested in multidisciplinary questions; we do also have non computer scientists among the authors, for example Prof. Nissenbaum http://www.nyu.edu/projects/nissenbaum/

Overall, as you suggest, our analysis is a mixture of CS, economics and social science.

Computer science _is_ a social science* -- especially when it comes to success and failure of a social network. You can't separate the software from the social and organizational context it's used in.

* more at http://achangeiscoming.net/docs/cssocsci.html

That's going to be a hard sell. I do research in programming languages now and I think you'd be hard pressed to argue that the fundamental nature of my work is a social science. Especially since the initiating paper included a proof of correctness...

Well, it's not _just_ a social science. There's still plenty of interesting work to do with 1's and 0's and proofs. But even in "hard CS" areas like programming languages, the social since aspects are pretty significant. What influences the success of new languages and evolutions of existing ones? Why haven't some obviously good featuers like pre- and post-conditions and invariants been better integrated by most mainstream languages? How to make functional programming accessible to a lot more people?

One dude putting a paper online arguing something doesn't make it true. CS is not a social science. At best, when you stretch definitions of 'within the field' enough, you can study things that are related to CS and have social implications, yes. That doesn't make CS a 'social science'.

While on the face of it such a study is closer to social science than anything else, the maths and stats involved in studying connections between people, computers and other man-made nodes, and even the communication of materials & information around the natural world, have massive overlaps.

It is one of those areas where the same maths is applicable to seeming disconnected practical/physical areas, like how psychologists are starting to use the maths of quantum probability to model certain human behaviours (apparently it seems to much more accurately model the fuzziness of our thought processes than any "traditional" statistics methods).

The applicability of the same maths to seemingly disconnected areas is one of the things that convinces many that a "grand theory of everything" may one day be possible rather than there being disparate base laws at different scales and locations (or the third possibility of course: turtles all the way down).

PS: If you knew what you are doing, you will never start:

Donald Knut said that if he knew the work that it would be when he started typography, he would never had done it( this means no TeX, and Latex, and art of CS books).

Steve Jobs and Wozniak said that if they knew what they were to do when they created Apple, they would had never started it.

About being "hard", I don't think it was harder for this twenty something to create facebook than other SN. He just did it right.

>There is the widespread — but often unstated and always unexamined — belief among the participants that moving to a decentralized setting is a magic cure-all for the problems that ail today's status quo, such as privacy and interoperability.

It's an improvement over data sitting in a central silo waiting to be mined and monitored. How would you suggest tackling the problem?

It's an improvement over data sitting in a central silo waiting to be mined and monitored.

I think this is a perfect example of the subtlety of the issues. Let's imagine that Diaspora succeeds and Facebook, MySpace, and Google+ all adopt its federation protocol; in that case it's likely that all of those companies would have access to most of what you post, effectively reducing privacy.

But they'd only each have the subset of your data that you choose to publish on their respective networks, right?

To give a specific example, if you friend someone on Facebook from Diaspora then Facebook would probably have cached copies of any data you publish that is accessible by that friend, including all your public data. (I guess you would have to agree to Facebook's TOS when you friend that person, even though you don't have a Facebook account.)

I feel like it's a form of opportunity cost that creates an imperials debt. Decentralization, and even more so all the motivations that lead developers down the path of decentralization, has nothing to do with making a good social network. It's like all the solar-powered cars... a cute idea, but no one ever (ever) buys a car for fuel efficiency. You buy a car to travel, to carry things. So the solar cars are cute, but they fail at everything a car is supposed to do. You can't possibly improve on something unless you start by implementing the thing.

I thought usenet would be a better starting point for how todo it.

And I thought email.

SMTP is not quite as close to the goal as NNTP.

"or do we tell him to learn some math and CS first, and gain an appreciation for why the problem is hard and probably not worth taking on unless you really really know what you're doing?"

The problem IS learning more math and CS, because odds are the people that make "yet another social network" already know about math and computer science as they use to be geeks.

So you are an academic and you believe they need more academics.

What they don't know is about marketing, aesthetics and art and psychology(things like Myers briggs PT) and all the multi facets that you need to know for a SOCIAL(people relationships) network.

People that love to program machines use to have a defined personality, the most difficult thing for them to understand is that there are other personalities as well that are different from them.

Actually, in the line you quoted, he's saying we need more academics (instead of amateurs) to solve the P =? NP problem, not to build decentralized social networks.

It was a metaphor.