Friday, January 23, 2009

Flow Language Basics

I wanted to summarize a fascinating conversation Eric Harris-Braun, Arthur Brock, Katin Imes, and I all had yesterday. Arthur and Eric have been developing the idea of a "flow language" for a while, but on the call we were all trying to make sense of what this new type of language is actually all about. The following is my crude attempt to describe the basics.

The purpose of a flow language is to enable the expression of flows in a system as opposed to discreet causal events. The standard form of language is subject / predicate. This structure makes differentiation between objects / actions very easy, but integration of those objects / actions into coherent patterns within wholes very difficult. Individual relationships can be easily isolated, but patterns in the whole are very difficult to express (or even perceive). A flow language makes these larger scale patterns visible and able to be engaged with efficiently.

The most basic “word” in a flow language is a currency. A currency allows one flow to be described. For instance, if we were going to describe the flow of water in an ecosystem, we would want to know how the water cycles through the ecosystem as a whole and how this flow affects the different parts. These parts are the core units, the relationships between which describe the flow in the system. Think of these units as accounts. A currency only exists as a series of relationships between accounts. Accounts are analogous to "letters" in the flow language since they make up the most basic unit, but they lack meaning in and of themselves.

Boundaries between accounts can be arbitrarily drawn. For instance, in the currency describing the water cycle, we could describe three accounts: land, ocean, sky. There are certain basic relationships between these accounts. We know that water does not flow from the ocean to the land unless it goes through the sky first. We know that the sky can rain water onto the ocean or the land. And that the land can transpire water to the sky or run off into the ocean. While this is a very crude mapping of the water cycle, the relationships between these three accounts do tell us something useful. We could easily get much more granularity by drawing account boundaries with more differentiation. For instance, we could draw a grid on the land, each square with its own account. This might allow us to see which parts of the land transpire the most water, and what parts allow it to run off. There is of course a series of relationships between these accounts that could be described (non-proximal squares of land can’t directly have water pass between them without going through the sky first). We could also draw account boundaries between species, or even between individual life forms. In any case, the account is the “letter,” and the currency is the “word” (describing the relationships between the accounts).

If we want a sentence in flow language, we must look at the interplay between currencies. To continue our metaphor, we might also look at the flow of nitrogen in our ecosystem. We could similarly divide the space into the necessary (and sensible) accounts, and examine the relationships between them. Perhaps where nitrogen accumulates, we see more plant life, which in turn allows the land there to hold more water and transpire it directly to the sky without running off to the ocean. It is these interrelations between currencies that allow for the full construction of a flow sentence. For now, I have been calling a “flow sentence” a “currency complex.” As dozens of currencies interplay with each other we see that more and more complex flow sentences can be constructed.


Katin Imes said...

What comes for me around this stuff is about scale and perspectives. It is viewing data from different perspectives on different scales that gives us the ability to see things "holoptically" and reveal many new things about the systems being viewed.

Scale means not only drill-down, which we do now and naturally, but also to chunk-up (NLP term). One question is: what chunking-up would be useful, and why? A concern around this is if we design a software/data system and it doesn't chunk-up well in the ways that we want, it not only will be costly to correct, it won't enable the emergent effects we are seeking to understand.

We might learn from the thought exercise of imagining that we are clicking on the word "stream" to drill-down... what other objects, relationships, flows, and relations between relations would it reveal? Trees, fish, rain input, nutrient flows and caches, and more. Now, what would a chunk-up from stream to bigger system reveal? What do you call the stream-sky-ocean complex as a single holoptic view (biome?)? Chunk-up again to continent-system view. What is useful to see there?

Applying this to money systems, what is useful to see at the chunked-up levels and what perspectives would we want to access or create at those kinds of scales?

Alan Rosenblith said...

To me the question of scalability is a question of where you draw the "accounts." If you want to think in terms of holons, an account is like one holon. For instance there is a series of flows that enable a tree that manifest as interactions between cells, but a series of flows that enable a cell that manifest as a series of interactions between organelles. The "account" is simply a way of specifying which level of holon you are dealing with.

In the simple example of ocean, land, sky, those are three giant holons that each have a number of flows within them, but the flows between them can be described in a simple currency. I think the ability to drill down or "chunk up" will have a lot to do with how accounts are mapped in an addressing scheme.

Katin Imes said...

The computation cost of drilling-down or chunking-up will definitely be related to how the data is structured. That's a main reason to try to understand the types of chunking-up views we'll want to see: to design structures that support those views.

Here's an idea for trying to "reach the next level": how about creating intermediate holons (in this case, summarized data sets) from the smaller ones. That would make computations of constantly-larger holons possible without demanding huge computing power and time.

When you are working with data such as transactions and trying to build holoptic views, there are no holons in between a single transaction and the holoptic view. If I want to see a graph that helps me understand the flow characteristics of all 'cw' currencies in the last 30 days, for example, I have no choice but to go out and find the thousands - maybe millions - of transactions that would qualify, and then total them up in whatever way gives me the perspective view I want. That's some serious computing power when the transactions are distributed all over the place. This computational cost makes scaling up and chunking-up costly.

What if we could create intermediate holons that summarized all the common perspectives of transaction sets. So there would be a holon already out there with the summarized figures for the cw.pdx currency, and another for the cw.sea currency, and so forth. Then all I'd have to do is add together dozens (or hundreds) of cw-summary holons to see my view. Once I had built my view, my large holon, it would be stored somewhere for others to use in building their views.

As long as we are going for the gestalt views, a few transactions that didn't make it into the data set for whatever reason wouldn't bother anything.

Alan Rosenblith said...

I like this idea very much. I was thinking that it could be part of a currency design to have clients send carbon copies of individual transaction records containing information about the transaction specified by the currency to some sort of aggregator. The aggregator could compile the relevant info and display it in some sort of holoptic view. This would get out of the sticky privacy question since the accounts partaking in these transactions wouldn't necessarily be specified (or only parts of the address would be specified to allow maximum ability to chunk up).

The aggregators could then easily themselves be aggregated for a larger holoptic view. This discussion might almost be worth starting another post for...

Anonymous said...

Integral Theory (Methodological Pluralism) addresses 8 indigenous perspectives by which ecological relations may be perceived. As integral theory is primed on holism-- holoptically is correct-- the proposition that currencies may be analogous to "flows" and accounts may augment certain subjects within an ecological topography but be valueless in terms of conventional accounting mechanisms. If general orienting principles are to prevail in absence of metaphysical moorings (if perceived out of value systems), may I suggest that the very Subject of currency/flow/value-exchange be at the discretion of a system of heuristic praxis. KAIROS: opening the moment of truth to its shadow and end.,M1

Anonymous said...

Hey guys -

ss here following the first anon with something you're reminding me of: a little UL practice/realization meme on the holoptic, which I associate with the transition to vision logic about which your discussion seems cognitively to pivot ...

Leslie Temple-Thurston (who's first book, "Marriage of Spirit" reads like a hacker's manual on the transformational process - or a sort of RoR for 2nd tier subjectivity), describes in another text the subtle experience like so:

"Let's look at this in terms of our schematic of the subtle body. Our consciousness naturally perceives ... unity consciousness at the third eye & above, at the crown ... Below the third eye, and down through the chakras, the ego perceives duality. This is where the third eye, the one single eye, located roughly between the eyebrows, becomes the two physical eyes ... is the origin of our either-or mind ..."

... which is fun to experiment with when flipping between/downloading from holoptic apperception to formop conceptualization.

Though areligious, she goes on to cite JC from Matthew 6:22, "If therefore thine eye be single, thy whole body shall be full of light."