The way I envision it, SocialNetworkSecurity is a cross-disciplinary field incorporating elements of SocialNetworkAnalysis & InformationSecurity. SNA is about understanding & exploiting the network properties of social systems; INFOSEC is about attacking & defending information systems. Combine the two & you try to find ways to attack & defend social systems by exploiting their network properties. You could think of it as a form of InformationWarfare?.

By whatever definition you care to use, I’m a hacker. I see the world in terms of attack & defense; discovery, vulnerability, exploitation & manipulation. As a hacker, SNA opens up a whole new world for me to explore & play with. Hacking is about finding predictable structures in black box systems & exploiting them in novel ways to suit my needs. SNA, which gives me a toolkit built around the principle of finding predictable structures in social systems, is a hacker’s wet dream. Mapping social networks is just the start; from there you can go on to finding key nodes that control the integrity of the system as a whole or the flow of information through it. Apply pressure (technological, personal, legal, media, political) on those nodes & the whole system suffers. Find ways to protect them & the system continues to function even under great stress.

I’ve held forth on this idea in other forums, see

Research on WA TeleCommunity for more of it & also some links that put them in context.

See also: nettraq on Citeulike

TimKeller 2006-10-06 04:44 UTC


Tim, this is an interesting idea, and I thank you for posting this.

The trolling community has studied this for ages, as have politicians and agents so on in the past.

I ran into an 80 year old man at a leftist bookshop, and he recommended to me that I subscribe to and write in to newspapers, posing as if I were a right-winger. Say embarrassing things, say things that are blatantly false, but believable that someone might think.

There are clearly weaknesses in the social fabric. But, I don’t think it’s totally defenseless. People seem to have more going on in their unconscious than I think most people tend to give credit to. People can instinctively feel when they’re being manipulated, and often react in ways to reject would-be infiltrators.

“Constant vigilance!”

The NetWar books is also an interesting starting point in this context.

Yep, I have that on my shelf at home. Another good history of NetworkWarfare? on the Internet is Wendy Grossman’s book net.wars. Moving into the more theoretical realm, Unrestricted Warfare (PDF), a book written by two Colonels in the Chinese PLA (which is IMO one of the most significant advances in the theory of war since Sun Tzu), touches on the subject in several places. And I’ll also recommend Richard Butler’s World Without Secrets: Business, Crime and Privacy in the Age of Ubiquitous Computing, although he does borrow other peoples’ ideas & name them after himself (most notably MetcalfesLaw). Other than that though, it’s a solid piece of work that outlines a decent set of guidelines for stabilizing social systems in a world of networks.

There are a number of papers on the algorithmic/SNA aspect of this, which I’ll be posting at another time (as an example, here’s one I just ran across: The topology of covert conflict}, but I want to draw attention to one non-technical discussion that perfectly maps to my idea: Understanding Jihadi Networks by Marc Sageman, M.D., Ph.D.

TimKeller 2006-10-17 02:12 UTC

Tim, I wonder if there might be some insights from SocialNetworkSecurity that could inform the some of the brainstorming that we are doing at MeatballAntiSpam, and SpammerSurvey? Particularly in the realm of SocialNetworkAnalysis, and social systems in general, as they relate to networks. Understanding the social networks of spammers, but using that understanding to try and devise ways to engage and understand their local life conditions and motivations, on both a personal scale (understanding the “node”/person) and on a network scale (understanding the social networks they are a part of).

So, in the case of the ideas at SpammerSurvey, for instance, we could look at behavior from a SocialNetworkSecurity perspective, yet the data produced is (possibly) used in the short term for “attack/defend” modes, which is more the goal of MeatballAntiSpam, but a related long term ultimate goal is to create both engagement with the people spamming, or at the very least, a valid and realiable knowledge base about their methods, motives, and as much information about their local conditions as can be obtained (more in line with steering and social change a la SpammerSurvey). And, of course, about the nature of social networks they are connected to. This is different than the internet network they are connected to. Their social network profiles would potentially shed light the nature of their social organization. So, we would be looking for ways to discern the nature of their roles in networks, and the nature of their connections, when possible. This is related to what we’ve been talking about at EcosystemOfNetworks.

We’d love to have your input there, if this is something that interests you.

I’ve read over the pages you pointed out & I’m game to join in because it’s clearly a worthy cause. But I’m stuck on finding a handle to start with. SocialNetworkAnalysis needs a source of transaction data to work with as the basis for building the maps that you can then analyze. Is there anything like that available regarding spammers? What visibility do we have into their social systems, that could be exploited for SNA?

TimKeller 2006-10-25 00:39 UTC

Thanks for those pointers Tim. 5GW is fun stuff that I’d never heard of. I really enjoy how they view it so objectively and coldly, in an almost gaming manner.

In the definition area of this page you say something like Apply pressure … find ways to protect them & the system continues to function even under great stress.

This is reminds me of what Ghandi (and Bal Tilak before him) called “Swadeshi” - which is economic sovereignty through local permanence.

Swadeshi is therefore a military defense tactic that utilizes the understanding of predictable structures in our physical reality - such as the fact that we must eat, and that all food comes from organisms, and that all living things need both land and water.

I’m no expert in such thinking, but understand this to be the larger reason that natural resources play such an important role in real-world military actions throughout the world.

I’ve been spending some time the last couple days playing the GPL {this builds cleanly for me, but let me know if you need help} which focuses a great deal on the importance of resource exploitation and pre-conditions to production. I’d love to mix this game with to show troops bombing water purification facilities, fertilizer plants, oil refineries, and even larger farms to weaken the ‘resistance’ to invasion.

Back to SocialNetworkSecurity: My point is that any kind of ‘security’ is enhanced when the community is able to stand on it’s own because it no longer fears sanctions (doesn’t need outside labor (new members) or extra resources to continue to exist), and cutting transport lines is not fatal (don’t know how to apply this one).

This is also related to describing 11-year old “needing jobs” because their society has been moved so far from Swadeshi that they are ‘lucky’ that they have any employment at all… But would employment be the goal if you were stranded on an island? It is clear that any job that can’t be automated away should be treated as a barrier to our real goal - to become self-sovereign or “set up” as described by Swadeshi.

Owners of productive sources can increase profits by decreasing the availability of alternatives. Because of this, eliminating competition, decreasing abundance, removing functionality or options and then charging more than ‘cost’ for this things are standard practice in business. We have become so used to these things that we will even defend the owner’s right to do so. And certainly the ‘right’ is there as part of property law.

But this is not the most efficient incarnation of production there could be. If productive sources were controlled (in the case of physical sources, I think simple ‘joint’ ownership is enough) by the very users of the objects of those sources, then there would never be any artificial scarcity. When I wanted to add power-windows to my car, I could do so without adding anything else, and it would only cost exactly as much as the extra materials and the extra labor to do so. If I held enough insurance, I could even do the labor myself by signing up for a time-slot to rent the extremely expensive tools in an automobile factory “at cost” - where “at cost” never includes something called ‘profit’, because there is no external owners to collect such a thing.

Owners will stop creating scarcity for profit by signing some kind of legal agreement (I’m thinking a lease) that causes consumers to become partial owners of the very sources of any object they use. One way to do this is to include a % of voting shares in the price of an object (say an apple) - so that the sources (the apple tree and all supporting sources under this lease) would incrementally come to be owned by those that use the objects of those sources. This will never be quite ‘perfect’, as there are always new object users that must initially buy at whatever price the current object owners are willing to sell at (how difficult is it to become part of this SocialNetwork), and old object users that tire of that kind of apple - so ownership must safely ‘decay’, but not too quickly…

The typical externalization of costs that keep current production limited in so many ways will finally be overcome in the physical world as is already being done in the virtual world because of the way the GPL requires object users be allowed “at cost” access to the sources of those objects. At that point we will be able to produce the vehicles and electronics we’ve always wanted to produce, but were not organized enough to do so without being hindered by owners that “prop up” the market to extract value by keeping those facilities artificially constrained.

I’m not entirely comfortable with your phrasing, it feels to me like it has roots in Communism, which IMO has been entirely discredited. However, I can see past that to what I believe is your central point, that by changing the structure of a social or economic system you can change its capacity for productivity & innovation. To support that idea I’ll point to this paper: Innovation flow through social networks: Productivity distribution. Summary: productivity & innovation within companies are directly related to their connectivity with other companies. I don’t have direct evidence at hand to support the claim but presumably the same principle holds true for internal connectivity as well as external.

TimKeller 2006-10-29 21:39 UTC

Tim, thanks for dropping that link in here. The idea in this is a key part of what I am trying to weave together with SocialSynergy:OpenValueNetworks?, which was inspired by VernaAllee’s work with Value Network Analysis, and the Communiting idea, and a lot of ideas from a lot of different people, some of wich is covered up in a recent paper by David Bollier, that I summarized for CooperationCommons? here:

I had not come across Innovation flow through social networks: Productivity distribution prior to reading your comment above here. Perhaps this is not directly related to “SocialNetworkSecurity” as an individual subject. But, I think that SocialNetworkSecurity is an important part of Socialsynergy:OpenValueNetworks?, and CollectiveProblemSolving, and P2P finance ideas like CommunityWikiBank, BarCampBank, and SocialSynergy:PeerInvest?.

Perhaps it isn’t & perhaps it is. I’d argue that ability to innovate is a major advantage in maintaining a system under stress or attack. The Iraqi insurgency is proving that their network structure is a major benefit to them offensively as well. Go to John Robb’s blog Global Guerrillas & read any of his posts on OpenSourceWarfare? for lots of proof of that.

TimKeller 2006-11-01 04:02 UTC

I agree Tim. I wrote about this here a while back : Globalization and Cultural Shifting

Yeah Sam, we’re definitely on the same wavelength there.

I’m starting to think about organizing some of the links we’ve been throwing up in the discussion here into a more coherent, structured format, but until I get around to that I’m just going to tack these two book pointers on: Annihilation from Within by Fred Ikle (with props to Global Guerrillas) & The J Curve: A New Way to Understand Why Nations Rise and Fall by Ian Bremmer.

I haven’t read Annihilation yet but from John’s review it sounds solid & lines up with my own philosophy & intentions quite well. As for The J Curve, its core theory is that if you make a graph with stability on one axis & openness on the other, you can plot a tilted J-shaped curve that predicts how nations react to increasing their openness from totalitarianism into democracy, moving from a highly stable but closed state rapidly into chaos & then much more slowly into regaining stability as the newly freed society learns to cope with its freedoms. It’s only a raw hunch but I believe it may be possible that complexity has a solution to describing the mechanism behind this, namely a cascade caused by phase change of some sort in the social structures of the society. In any event it’s a fascinating & quick read, worth the investment IMO.

TimKeller 2006-11-09 15:56 UTC

Wow, Awesome. I will definitely read. Please let me know when you start turning our discussion here into DocumentMode, and I’ll be sure to jump in and help you. Also, we may eventually spin this off into it’s own wiki, if there is interest in this (of course, we an easily connect in a WikiNet to Community and Meatball, CollectiveProblemSolving Wiki and others.)

Tim, thought you might find this interesting in the context fo SocialNetworkSecurity: The Shifting Hub. Which makes me wonder, if there is a way to harness this shifting hub effect in the way that people are solving problems, to make social networks more secure from information cascades, and echo chambers that create and drive them?

Yeah I saw this when it first came out, it’s a significant advance in understanding how social networks work. To summarize, there’s two separate topologies at work in a social network. First there’s the structural, relatively static topology of who knows who, and then overlaid on top of that there’s a more dynamic transactional topology of who actually talked to whom within a given span of time. A lot of time’s been spent looking at the first type of topology but less effort’s been put into the second, which is where I think the real meat of driving change through SNA is going to be made. Although they’re clearly related (if only because I can’t talk to you if I don’t know you), our ultimate objective has to be change in the patterns of actual communications & not just the static structure of the web of relationships between us. To accomplish that we need tools that can help distinguish between good ideas & bad ones so we can amplify the good ones & tamp down the bad ones. I strongly suspect reputation systems can play a major role in that regard, although I expect there’ll be a lot of resistance to them. For a foreshadowing of what I’m talking about take a look at the initial reaction to Clay Shirky’s Power Laws & Inequality paper; many were horrified at the idea that some people had more influence within a social network than others, decrying how unfair it was & demanding fairness for all. I fully expect there’ll be another round of the same when reputation-driven discussion & decision making systems come on line. Another type of hypothetical tool, more specifically in the direction of what you were driving at, would be able to recognize degenerate patterns such as information cascades within an ongoing conversation & alert either the participants or some third party who could then intervene to break the chain.

Welcome back, Tim! I think you are right about the resistance to reputation systems. Although, at the same time, in CommunityOfPractice situations, where people are actively trying to apply ideas, there is already a filtering criterea for ideas, usually. Many networks of collaborators have these criteria in some form.

For example, if you look at BarnRaisingNominations on MeatballWiki, you can see a kind of CommunityFiltering? “Peerreview” of this group activity. I think it is more acceptable to some people to have their ideas at least considered, and judged on the merit of their idea, rather than based on who it came from. I think this can be a “flip side” of the Power Laws & Inequality: some groups of people will give more credibility to ideas that come from certain people because they have, for instance a higher ranking in technorati “authority”.

This “authority” really only means a bunch of people linked to them, in the case of blogosphere “PowerLaw”. But, it doesn’t mean that truly knowledgeable people have helped to PeerReview an idea, necessarily. This is where communities of people interested in an idea can come into play, so long as they are able to create a fairly objective system of PeerReview, this PeerReview can actually be the basis for quite an acceptable “ReputationSystem?”, at least to most people.

I actually think different communities, who are based around different fundamental assumptions of what they see as “valuable”, will end up having different “ReputationSystems?” emerging around their activities (see LiteracyOfHumanNature for a discussion of value systems).

I think we’re talking about two different types of systems. While what you describe can definitely work at the local level of a CommunityOfPractice or WorkingGroup? where the participants know each other well enough to organically filter out bad ideas based on their merits, that’s not going to scale up to a global enterprise-level system with thousands or even millions of members. You can’t put every idea of that many people up for open discussion by the entire membership of the group. You need a system that pre-selects the best ideas & also pre-selects the best people to sponsor them.

For some rudimentary examples consider SlashDot’s Karma system & DailyKos?’s DiaryRescue? system. In both systems every member gets to contribute his thoughts to be read by anyone who chooses to drill down for them, but only those that are considered worthwhile by someone who has a significant reputation within the system & is willing to impart some of it for the sake of a comment or post get elevated to the level of global visibility.

I’m only dipping in from a remote place, so I don’t know everything you’re talking about.

But I have an idea that might help:

  • Look for “different” ideas, rather than trying to isolate the “good” ones.
  • Assist the emergence & creation of maps of discussion spaces.

Tim, yeah, I think we were talking about different scales. EcosystemOfNetworks describes the different scales we are talking about, I suppose. There are also certain criteria that content is filtered on in WikiPedia that are on the scale you are talking about, too.

One of the things that can scale a bit beyond working groups is ConsensusPolling. Check out this is kind of a like a bridge between the smaller collaboration group, and the more stigmergic, PolitcalSoftware?, larger scale that you mention above. ConsensusPolling adds some extra dimensions (not direct reputation, but a process of PeerReview).

I think some of the reputation problems might be solved by being associated with a “group”. Because we currently lack the time and accurate metrics to create universal reputation systems outside of controlled enviornments (although, some of the ideas like are coming closer to a more natural, and likely to be used system). For ideas, most useful, actual work is done by smaller groups. Discussion, or “crowdsourced” highly distributed “mechanical turk”-type work is what you usually find on the larger scales you are talking about, online.

That doesn’t mean that sites like SlashDot, DailyKos?, Digg, and other larger scale sites are not important or influential, because of course they very much are in many ways. But, there are some activities that I am finding myself participating in, online and offline, among networks of people, that are involving more intensive collaboration on smaller number-of-people scales. In the case of these smaller groups, they can “scale” to enterprise levels, when they join together in loosely-coupled ways to solve problems. This dynamic is different than the same amount of people who are largely anaonymous to one-another visting a site like Digg or SlashDot, or read/comment/rate. One example is a SocialPublishing project like This is reading, commenting, and rating, but also producing, editing, etc. WikiTravel also comes to mind.

Basically, there are lots of dynamics online beyond blogs and SlashDot types of sites. I was once very excited about ReputationSystems?, but I now realize the reason why we don’t see them everywhere is because they are not universally useful in all communities. Nor is scaling up to enterprise level:

I don’t think this means your are “wrong”,Tim, I just think there are different dynamics at different scales, and I think we are starting to learn how some processes work better on different scales.

Lion, maybe you can expand on what you are talking about?

I was responding to:

To accomplish that we need tools that can help distinguish between good ideas & bad ones so we can amplify the good ones & tamp down the bad ones. I strongly suspect reputation systems can play a major role in that regard, although I expect there’ll be a lot of resistance to them.

The signaling theory sheds quite a bit of light on the subject of security in social networks.

(..slipping by Zby)

To accomplish that we need tools that can help distinguish between good ideas & bad ones so we can amplify the good ones & tamp down the bad ones. I strongly suspect reputation systems can play a major role in that regard, although I expect there’ll be a lot of resistance to them.

Lion, I see. I hoped that you might outline some form of “vision” for the type of UniqueIdeaIdentifier? that you were talking about above. How might this work?

Oh, I know the answer to that one! I have a whole plan and schematic for it, … …in my head.

You want a tagging system that allows you to label the parts of an EmergentStructure, and then contextualize them by domain, into an AggregatedVisualDisplay for MappingTheIntangible, with SemanticRegions?. You can use TagsForTags to help identify ConceptualFocalPoints. It’s the AbstractionIdentificationProcess.

I know how to do this; I just need to take the time out to write it all up, with diagrams and pictures and stuff. I already have diagrams on 3x5 cards.

I am confident that it can work individually (by yourself,) collaboratively (with others, intentionally,) and incidentally (with others, accidentally, not intentionally.)

I’m busy writing a paper right now, so I can’t make the project sheets or CommunityWiki pages for it just yet. But there is a clear answer in sight. I’m convinced this (or something very much like it) is an essential part of HiveMind visualizations, and that something like it is a transcendent step. VisualWiki is a good part of the way towards this.

I think I also see more clearly what Tim is referring to above. I think he is saying (written here in my words) that in very large online communities, ReputationSystems? might be one component to help stop InformationCascade? actions on inaccurate information, distortions, PropaGanda?, etc.

We can see this in sources like NewsTrust, which is a very large scale collaborative filtering of news. It does not yet have direct reputation of participants, but it is a step in the direction that I think TimKeller is talking about. The idea of SocialNetworkSecurity being to stop people from harnessing the power of very large scale SocialNetwork and information systems for nefarious puposes, like spreading fear, manipulating people on a large scale, etc. I see Tim’s reference to SlashDot CrapFilter? as a kind of filtering that could be done based on reputation.

I think that we are nearing an applicable set of tools that will let people start to choose many voluntarily offered reputations to view people through. Also, PortableData? will let people start to look at the ratings of content created by people, as well.

Zby, that Signals paper is awesome, I’m going to have to take some time to digest it. In the meantime, let me introduce StupidFilter, a Bayesian filter that weeds out stupid comments from blogs. It doesn’t directly use any of the concepts we’ve talked about but it is an attempt to achieve the same goals.

Tim, I think you might be interested in this paper, too: SocioCyberneticDecisionMaking? PDF

OK, here’s the first tool I’ve seen that incorporates the concepts that make up this page. It’s called Organizational Risk Analyzer. Basically it lets you analyze the topology of a given network from a security/risk viewpoint, identify critical nodes within it & use a variety of algorithms & techniques to project how changes to them (loss of function or compromise for instance) would affect the network as a whole. Very cool stuff.[1]

Three links on the same subject, a demonstration of SocialNetworkSecurity in the real world. This strongly suggests that in order to effect a victory against this type of movement, one must use the concepts laid out on this page to attack the structural integrity of the group/movement as a whole.

The Mathematical Structure of Terrorism

The Mother (Nature) of All Wars? Modern Wars, Global Terrorism, and Complexity Science

Universal patterns underlying ongoing wars and terrorism (the original paper)



Define external redirect: CommunityFiltering PolitcalSoftware SocioCyberneticDecisionMaking InformationCascade PeerInvest OpenSourceWarfare CooperationCommons NetworkWarfare PropaGanda PortableData DiaryRescue InformationWarfare ReputationSystems FifthGenerationWarfare CrapFilter ReputationSystem OpenValueNetworks WorkingGroup DailyKos SemanticRegions UniqueIdeaIdentifier

EditNearLinks: TimKeller BarnRaisingNominations CommunityOfPractice BarCampBank VernaAllee SlashDot MetcalfesLaw MeatballAntiSpam WikiNet WikiTravel InformationSecurity DocumentMode ConsensusPolling PeerReview NetWar MeatballWiki