- for future uses?

Lately some of us are increasingly interested in tools for sense-making in our online communities (as with Edgesense and Assembl, or OpenEthnographer), so I just wanted to flag Metamaps as something that looks interesting for the future, or who knows. It’s free and open source (though in private beta), and it seems complementary to Edgesense which looks at the network graph based on the original database, and Assembl, helping us work with content in a friendly, albeit less robust way compared to Open Ethnographer.

It might be worth looking into if we ever decide we need a quick way to visualize Edgeryders projects, groups or subcommunities and relationships between them, particularly for newcomers who might want to get the overview right away. And this could be the case given that we’re increasingly becoming part of a network of networks, as some like @elf_Pavlik would argue (correct me if I’m wrong).

Intro to from notthisbody on Vimeo.

I know folks in Sensorica, OuiShare and Pavlik himself have been using it, so we know who to ask for advice when needed.


What do you like about it, Noemi? I don’t get it at all sad. It’s not argument mapping. It’s not a social network representation. None of the featured maps has more than 5 contributors (the one with 5 contributors is about Metamaps itself). I have browsed a few, but can’t make sense of them. It seems they are using the same representational logic of a graph database: any two entity of any kind might be connected by an edge, because an edge can mean anything.

Alberto => Noemi

“=>” means "is commenting in this thread

Alberto => a nice tomato salad

“=>” means “has eaten yesterday for dinner”

And so on. Graph databases think in triples: they specify the two entities being connected and the meaning of each individual connection, and it all works as a way to do semantics with computers.

But if humans see different relationships represented in the same way, they are likely to get it badly wrong. It’s just bad visualization, paving the way for misunderstandings and non-rigorous thinking. People who do semantics with computers do not have that problem, but then they appear to spend an inordinate amount of time trying to pin down just what exactly we should call different entities and their relationships (I loved this post about the wrangling of it all).


I work with metamaps project, and do quite a bit of mapping there. I do encourage experimentation by anyone interested in the arts of systems thinking, collaborative sense-making, collective intelligence, networks, etc.

An important distinction to make is that metamaps is very “hands on” and participatory - there are relatively few algorithms at work here…yet. You create the map as a representation of complex subject matter, and then work with it any way you like to help organize and understand, iterate and refine, expand and subdivide. Your peers can do the same, on the same map or a fork of it. Topics are shared in the entire platform database and as you are mapping you are prompted to reuse existing topics where appropriate so as to weave an increasingly detailed contextual web around significant entries, which can be navigated through a separate portal called “topic view”.

For example, the EdgeRyders topic “graph” on metamaps looks like this at the moment:

…showing connected topics from any map in the system which contains that Edgeryders topic. You can right click any of those topics in turn and “get siblings” to see what’s connected to them. This is a great way to quickly explore the collective “knowledge graph” - and if desired, save what you see to a new map and continue adding to it there.

Alberto, connections on the map (aka “synapses”) in many cases do have a label or type - when creating any connection the mapper is given the option of labeling it with a word / phrase of their choice. Both topics and synapses can be sorted and filtered by type on the map view. To view the synapse label / type you need to hover or select it (click or shift+click+drag).

Currently, data is accessible in JSON format and we’re working on a more advanced API which will allow convenient import and export in various formats and the potential to interoperate with other instances of metamaps and similar LOD information architectures.

I love what metamaps has done for us as a peer network collaborating to develop and distribute the platform - a powerful cognitive leap for individuals and communities of practice. I like to think of it as a “pattern language operating system”. It’s kind of fun and pretty too :slight_smile:

Happy to answer q’s or lend support for anyone interested in learning more or trying it out! Thanks for expressing your curiosity, Noemi.

Triples for humans?

Hello @BenB, welcome!

Aha, I see. You are trying to build a sort of triples store for humans! Maybe it’s just me, but don’t you run the risk of it being neither as rigorous as RDF-type relationships, with their tight specifications, nor as intuitive as standard-fare graph representation, with only one type of nodes and edges and straightforward interpretation?

But then again, we all have our pet tools and it not for me to question what others like! smiley Plus, I am a very very poor visual thinker, close to cognitive impairment in fact, so I am possibly the worst person to use (visual) pattern language stuff.

Is it intended for stewards/ network weavers or for all?

Thanks so much for the clarifications @BenB! What drew my attention to the tool was the way it can be used to build a network visualization that doesn’t rely or retrieve data entirely on/from one community network graph, but includes outward connections of that community eg projects that may not happen on the same platform. So I spotted it as a potentially useful tool for community builders first and foremost - meaning translating my own mental visualization and invisible knowledge into a useful resource for newcomers, not pretending a lot of people would use it and develop it further.

I see you yourself have done some work mapping Edgeryders-  upon using the Topic search I came across this LOTE3 map, so thanks :slight_smile:

Now, my only doubts - take it as a very early feedback - are in terms of:

  1. being massively participatory: do you have any proof that people are motivated to fill in the blanks by editing someone else’s map? Because of the subjective meanings of the relationships between elements, I find it difficult to jump in and contribute to an existing map. I’d rather start a new one and follow my own thought process. But the what you end up with is a lot of maps that are used very little. Not sure if that’s a problem to you as all the data is anyway aggregated in the system (and maybe that’s your main goal), but here at ER we have a similar problem with the Projects listed by contributors, which follow the power law distribution pattern present in most online networks. Very few turn out to engage more than the initial contributor(s), unless a big help is provided by the minority stewarding the platform. And that is problematic, because what happens is that when new people show up, they most likely won’t be joining projects that are harder to grasp or find because the system buries them to make room for the new ones.

  2. scaling: how does one make sense of an already complex map (eg Metamaps Ops) that will get increasingly bigger? The solid networks will keep on growing, and even if the core people in the network will strive to map every new relationship, it can still become impossible for newcomers or outsiders to make sense of it, therefore defeating the purpose. And that may not depend on how much of a visual thinker you are, it could just become practically illegible… or not?

Great thinking

As always, @Noemi is more eloquent than I am.

I guess her point 1 is why semantic web people bicker so much about painstakingly precise definitions of things, like so many medieval scholars. Without a shared meaning, other people’s maps are unreadable (not just Metamaps, the same holds for paper maps of physical territories). And I guess her point 2 is why triples stores are usually meant to be machine-read, not humans-read. Well-defined ontologies and massive computing power make even very large graphs interpretable, at least that’s the theory.

Maybe Metamaps are meant to be built rather than used – like Post-its, or Ola’s decks of cards. It’s also what you see in the video: people around a table, trying to make sense of some vastly complex thinking task and smiling as they “get it”. The people around that table are probably going to hold enough shared knowledge that they can make sense of the map they built, but that knowledge is tacit, hard to share outside the group. In I come, I look at the map I have not helped to build, I just can’t make sense of it. In fact, the way I think about networks is in the way: I experience a sort of buzz in my head as some of my subroutines try to superimpose one-mode networks onto what is, patently, not one such.

Connection points

Something to consider in approaching an unfamiliar map is perhaps first identify our connection point to it. Is it in a subject domain that we know about? Was it created by someone we’re connected to? Do I recognize some topics on the map that are relevant / of interest? Does the description give a clue about the purpose of the authors? Basically, that’s the process of discovering the latent links between your own working “map” (point of reference or line of inquiry) and what you see in front of you on the canvas which was created by some peers. If there is no obvious connection, then most likely you pass on by and forget it. But if at least some of the topics are relevant, then perhaps you start there and see what is connected to them on that map, or else deeper in the system (the “topic view” approach mentioned above). You don’t need to process or comprehend the whole thing at once to interact and find value there. Although, it’s certainly nice when that happens easily, which is part of the skill in building really great maps for the purposes of presentation and broad engagement. Same with authoring a blog post, or an infographic, or…most any media.

I think we are starting to see different “modes” of mapping, some of which are very divergent and exploratory (and hence probably hard to parse for a visitor - like a “brainstorm”) and some which are very clean, elegant, convergent, and generic, like a refined diagram to illustrate a particular concept, narrative, or dynamic. Soon we’ll have a better implementation of “favorites” leading to “featured maps” which should reveal more of the latter, up front for curious visitors.

Appreciate the considerations you raise

Indeed, you could say that metamaps is a kind of ‘curated’ or ‘hand-crafted’ network graph :slight_smile:

In the immediate case of one map, that means it is often relatively personal and contextual. However, the slow accumulation and versioning that can happen when many people from a community or network begin to create and connect their individual maps, frameworks, and perspectives is a very informative and enabling process. So yes, to answer your question there are some (but not many, as of yet) instances of substantial collaboration on individual maps, but what is currently a bit more difficult to see on the surface is that there are many cases where people take inspiration from or make a direct copy of one person’s map and then go to work on creating their own version of it, maintaining some connections beneath the surface to those of peers through the use of recurring topics and structures.

One upcoming feature that we hope will greatly enhance the experience of live collaboration on shared commons maps is the addition of audio, video, and chat communication channels right on the map canvas. Already, if you’re on a commons map with another logged in user you’ll see their avatar icon moving around in real time and any edits they make.

Another point to your concern #1: something important about the information architecture in a map / graph based system is the  accessibility of relevant content in a much more context-aware fashion. In the case of more traditional databases, taxonomies, and file structures, content lives in one “official” location (say, a project space, to use your example) and is not readily visible to someone working outside that space (a new or different project) unless they know of the existence of the former, or can see a clear relationship between them which has been defined at the system level (say, a hashtag, or higher level category such as “projects relating to social enterprise incubation”). In a graph database, you start to see lots of emergent pathways linking content and concepts, which are user defined, and are accessible in various ways to suit the preferences, intentions, or conceptual framework of other users. It takes some getting used to…but ‘surfing’ a mature collaborative knowledge graph environment is pretty empowering and intuitive!

One quick note to your question #2 is that once the information is there in graph form on any map, it becomes easy to process and package in more presentable or purposeful form. For example, I actually made a ‘fork’ of that Ops map a while ago exactly for the reason you bring up - something that would be more digestible and relevant to the ‘average’ guest user. Copied the map, removed some superfluous information, rearranged it, and saved as “Metamaps Ops Public” here…

We need to find a better way of showing the sorts of connections happening between maps like a Git fork-diff-merge versioning system of some sort…for now, you can at least place one map as a ‘topic’ onto another map, with the url link in that topic card to jump between. Lots to build, test, and explore! Thanks for taking the time to investigate and share your impressions! This is a fascinating aspect of open collaborative networks and ecosystems right now, and I’m following your work on CAPS with interest.


No, thank you Ben for such an eloquent reply, it’s fascinating how much deep knowledge goes into building imperfect tools. It’s the same with Edgeryders too, and what’s even more valuable I think is the processes informing the many options we now have to move on to more mature collaboration within / between networks. I think you guys are onto something, although I barely use a couple of web tools so can’t really compare.

Again, working out the processes into something that’s relevant to critical numbers of people - therefore building significance - is what some of us strive for. And that is probably harder than building the tool itself… so the more we get familiar with one another’s ways the better.

PS True, your new version of the Ops map is much better :slight_smile:


Yes, semantic rigor is not the foremost aim in this approach to mapping knowledge. Rather than pseudo-objective it is I would say intersubjective, providing the user the chance to create their own personalized maps and models in a way that still connects to a larger emergent picture crafted with peers to the degree they wish.

I think an important part of the process (and novelty) on metamaps is that it becomes a space to explore and evolve and interweave categorical systems and models, such that multiple perspectives / frames can coexist and interoperate, and ultimately reveal deeper patterns.

Right now the system is constrained to a predefined set of ‘native’ metacode (icon) types for roughly categorizing types of topics, but ultimately we’d like to see it open up to an API with The Noun Project and other collections of ideographic symbols, going a step further towards language-agnosticism and what I think would be a deeply intuitive system for representing non-linear meaning and narrative.

Imagine if some maps are not just a representation of linked data / info, but actually a model of the operative knowledge architecture itself (metadata structure) and its relation to other models in use on the platform…showing relationships between categories in a flexible user defined way, such that I can map how my categories relate to your categories, and navigate your ‘information space’ and data sets on my own terms rather than some generic and rigorous standard. You start to get into a visually programmable p2p information ecosystem, with implications for a bunch of interesting challenges that arise on the way to a decentralized semantic web.

But that’s all kind of far out stuff…right now, I would say the most interesting and accessible applications for metamaps are things like system / network / org mapping, design thinking, participatory sense-making. We’ve got a reasonably active G+ community to check in on for more ideas and use cases:

Visual thinking isn’t an easy or natural switch to make for people (most of us) deeply enmeshed in linear cognition and media…but I think it’s a crucial literacy to explore and practice for the information systems of the future! It can be fun too; game-ful, in the right circumstances. Enjoy!

interesting challenges

“You start to get into a visually programmable p2p information ecosystem, with implications for a bunch of interesting challenges that arise on the way to a decentralized semantic web.”

As I mentionned in a post :

and a further comment :

I have been thinking a lot about that, and I think I have a plausible development path; but the challenge I have not yet solved is :

“how are we going to prevent that the tool we develop will be used mainly to perpetuate hierarchical power structures ?”

I am looking for the best combination of tools and communities to have that discussion.

"Visual thinking isn’t an easy or natural switch to make for people (most of us) deeply enmeshed in linear cognition and media…but I think it’s a crucial literacy to explore and practice for the information systems of the future! It can be fun too; game-ful, in the right circumstances. "

Since a long time my internal way of thinking is mainly “visual” network (graph) oriented, not linear. My problem is having a tool to be able to communicate that way with others.