This is just a quick note about a dreadful news: IT business is running short of trigrams. Trigrams are those nasty special words made of the three capital letters of some difficult to understand expression we traditionally use to attract interests in a new Graal. Running short of such beasts is a nightmar because without trigrams, we won’t be able to communicate about our new fashionable things…
Lately I re-read a list of errors about computers [fr+en] (sorry for English readers, items are in English but comments are in French) found in movies. It’s not exactly the first time I’ve read such a list and it’s always amusing to see a collection of problems movie makers do about our domain, but this time something popped up in my mind: a lot of those errors are not errors at all, they are what we, as common users, would request our computers to do for us.
Let’s take some examples:
Isn’t it a dream for all of us to be able to interact with a computer without the hesitations, mistakes and inversions that we do every day in speech? Is it a so terrible thing to request that computers should be able to understand little errors and avoid bad misunderstanding? For example when a compiler says: “Error line 10, missing termination quote.”, is it bad to think that it should be able to add it without complaining?
Well… Obvious isn’t it?
Why should we still have to bother with backups? Computers are known to fail for a huge collection of reasons. Some of the failures may even appear as pure expressions of the chaos theory (same causes bring different consequences). Why aren’t they sold with built-in gear to protect against such unpredictable behaviors? Is it that the new stuff would be also candidate for pure epidemic failures?
Is it so laughable to imagine a computer that would not hog all resources you give it? Is it so unrealistic to imagine a computer that does not spend a precious time to do strange things as waiting tenth of seconds for an electronic device functioning at light speed to finally conclude that the device is not present?
Standards. I pray for standards every day. Is it surprising for not IT-trained people to have difficulties to understand why each new thing governing the IT world exposes a new completely re-designed and non-compliant way of doing things? As a specialist, I do not understand either.
Ok, they laugh at movie-makers because they don’t know, those poor guys, how computers go. But wait! What if we were the real culprits because we are not able after all these years to built computers that act correctly?
Here and everywhere
Here in France we have to face a very dangerous bill [fr] that may break some fundamental principles of the internet. It is a quick and dirty answer to those who claim that exchanging files on the net would lead (and in fact would have led) to big losses for music and movie major companies: the solution is called the graduate riposte and authorizes non-juridical organizations to arbitrary close personal connexions and sue users without any proof.
This is an example of an un-designed solution. Rather, it is an epidermic reaction to profound changes in the way human culture pieces are both created and transmitted. Instead of designing new business models, old companies try to reproduce in the digital world what used to work in the physical one. And to achieve this out-of-reach goal, they consider all exchanges as a priori unauthorized, making all people pirates and felons. However, it is particularly astonishing, because a lot of brilliant researchers have written and talked about new business models and some companies have even tried some with success. Even some musicians have deployed new strategies to increase both their fans number and their incomes by using new principles of the digital age.
How did they proceed ?
By carefully analyzing both the fundamental being of their creations (digital work) and the different ways they can use to push their work to people that could be interested in it, they have been able to design exchange protocols and new offers to reach the correct compromises to sell their work, while in the same time increasing the number of people ready to buy it. They have found that far from creating losses, the exchange of music between teens is an incredible way of increasing the awareness of their production, leading to an increase of their sells, provided that the music is available easily and that the bought song can be played on any of the devices a buyer may use (which makes DRM systems a very bad thing to waste money on).
Fundamental being of digital creations and transmitting protocols between humans using internet. Sounds like “semantic” and “pragmatic” studies, doesn’t it ?
To learn more about how and why Praxeme separates these aspects read first this paper: and then dive into the Praxeme Institute web site.
In the open source community we see very efficient results emerging from non organized groups of people acting individually. Companies should fear this mechanism because it cannot be used inside organized entities. There are economical explanations of this for example in this book from Clay Shirky: Here Comes Everybody: The Power of Organizing Without Organizations reviewed in this article from Bruce Schneier. Here I want to focus on another aspect of the problem: emergence is an evolutionary mechanism.
Ants and other so-called social insects have long posed (and continue to pose) problems to biologists, ethologists, sociologists, and a lot of other disciplines: how could so complex behaviors and capabilities be hosted in a so small brain? How could a so small animal be able to build sophisticated constructs, with thermal mechanism, mazes, defense buildings? How optimized paths to food could be found with so few neurons?
The main answer to those questions is emergence. We now know that ants and termites as individuals have relatively simple behaviors; They respond to a limited set of simple rules. For example, when an ant gets out of the anthill for food it first do it randomly. While traveling it leaves traces of pheromones on its way. Simple behavior. If it finds food, it then follows the traces back and adds some more pheromones on the path. When another ant gets out of the hill, it has a very simple rule encouraging it to follow a path of pheromones. The most pheromones make the most attractive path. So a path with a return (two traces) is more attracting than a path with only one (no return, so no food chance). And a new ant will add its trace, going and returning with food. Statistically, paths to food will survive in the set of all possible paths, until the food resource is exhausted. At this point the path will begin to loose attraction because ants won’t come back on it very often, searching randomly new resources from the exhausted one location.
Sometimes, more than one path lead to the same food resource. The shortest, in time, sees more ants walking on it, more pheromone traces added to it. The concentration of pheromones on this path increases hence the path becomes more and more attractive compared to other less walked ones. In a short time all other paths are abandoned and only the shortest (in time) will survive. Ants have chosen the optimized path. How?
No ant is able to compute the optimized path. They even cannot compute a path at all. In fact it is probable they don’t know anything about paths and what a target may be. The choice is only a guise. A choice implies a conscious act and no ant is able to do something like that. The concept disguised as a choice is called emergence: from a collection of simple behaviors created through simple rules, a complex behavior emerge, which is the product of a large number of simple programmed actions instead of a complex computed and decided strategy.
Emergence shows its face in human behaviors too. The financial market is a very complex domain. We know that nobody is able to understand it and predict its behavior with great accuracy. Not to say that nobody has been able to build the financial market from scratch. Though there is something complex, human build, observable and it is called financial market. It emerged from a huge collection of individual actions made by a lot of humans. It has its own internal and unknown logic and may be seen as a homogeneous behavior of a big entity (the Market) with a virtual consciousness like the ant heap may be seen as an organization with virtual consciousness.
Our IT systems are today like the financial market: the complex result of a number of individual actions ran nearly randomly out of any strategy. Each move adds its trace on the pile of pheromone: it has been used here (and no matter if it has been used for good or for bad). And the stronger the pheromone track, the stronger the urge to act like others. Could something as efficient as an anthill emerge from this chaos inside companies? No chance!
Ants and anthills have evolved during a very very long period of time with a huge number of attempts (generations) before the result we see today. Amongst this huge number of tries a lot did not have the slightest chance to survive. In fact the vast majority were not viable at all. In open IT communities, we see the same mechanism: the number of actors is potentially huge and they all act according to there personal skills. Ideas and projects may be abandoned without consequences and the number of tries may be important enough to induce a good emergence phenomenon. But for a company IS we don’t have opportunities to get the number of tries to see a correct solution emerge from the chaos. No time. With the very few number of tries we can count on, there is no chance to even see a viable solution. Not even the tiniest hope. This is just a cold statistical computation. Period.
Emergence is not the way but methodological building is. There is a huge difference between nature and open source communities in one hand and companies in the other: goal. There is no goal in nature, there are precise goals in companies. Goals must obviously constrain all company actions. This is the essence of a company as an organized entity. All actions not aligned with goals are lost and waste part of the limited amount of energy available. The only way to have all actions aligned with goals is to deduce actions from goals by conscious decisions. We as humans are the first (on earth) entity to be able to build things by a conscious act of will. We have the ability to analyze, criticize and built new answers by deciding to examine problems and deduce solutions. Why not use this capability?
The Catalog syndrome
The Catalog Syndrome is this special way of solving problems with only technology: we have a technology so no matter what the problem really was, it’s solved. Why spend a lot of time in analysis and thinkings about a problem when we can jump in a technology which is obviously made to solve it? More and more often our first response to a problem is a technology. And in a lot of cases this is the only response. Technologies are chosen from lists aggregated in big families called “environments”, hence the term Catalog syndrome. Take a number of example:
During the last decade I kept asking myself: where does this technocratic tendency come from? Is this inevitable? I think some researchers in other domains have answers. Perhaps not ultimate answers but surely some hints to understand how we as humans may have come to such behavior.
Replicator
Darwinism theory states that evolution works with three main principles:
So genes are called replicators. And with the three previous principles we have an abstraction of what a replicator is.
Is there another replicator?
Think about ideas. And stories, songs, legends, theories, fashions… All this things are transmitted. One person passes them to others by a number of mechanisms, basically: the emitter duplicates the information from his own mind to an intermediate media (writing a book, making a photography, performing a song on stage, telling a story, discussing…). The media is then used by targets to reproduce the information in their own mind. Each information element is reproduced from mind to mind. The information element is called a meme (this is a strange word for French people, because it is very near the word “même” which means identical). Learning, earing, seeing are ways for our brain to start the imitation process. So memes are copied inside one brain and from brain to brain by imitation.
Guess what: copying involves variations. There are a number of mechanisms to explain variations in copying memes: bad earing, misunderstanding, bad singers, creation desires and innovation, bad memory, interactions between memes (mixing, replacing, cutting…) and so on. All these mechanisms take place in our brains where the memes live.
Our brain has a limitation on the number of memes it can hold. Despite the fact that we can think about a lot of things (using the “discursive” thinking for example), we cannot embrace all ideas and information elements that our brain may be in contact with. So memes can be seen as fighting to survive in brains. Some memes are more compliant than others with the context made of all memes leaving in a brain at one time and so they will have more chances to survive (be reproduced). Brain is the natural selection context where memes are selected according to survival capabilities.
So we have:
Hence memes are replicators.
It has been shown that it is very probable for memes to have an impact on genes. But it is another story.
Memeplexes
Genes are arranged in complex groups with strong links between them and heavy interactions. This is shown by links between elements of the phenotype. By grouping, genes can create complexes with an increased robustness that benefit to all the intervening genes. In the same way we see groups of memes that are more robust together than the separate ones. A theory, for example, which is a very complex group of memes, has a lot more chances to survive than its components. Even if the theory is false (proved false by another complex group of memes). Or a faith supported by a long history of believers and reference books. Complex groups of memes are called memeplexes.
As we have seen, memes and memeplexes reproduce by imitation. So a robust memeplexe should be easier to imitate than non robust ones. An easier to imitate meme will have more chances to survive in one brain and to pass from brain to brain (It’s out of scope to define what “easy to imitate” could be).
Back to the beginning
Just compare two things:
Which one is easier to reproduce? The pure idea has to be read, understood, explained, rewritten in other words or contexts. On the other hand a technology is there, with the idea hidden in it, and with apparent results and benefits, directly, without understandings, explanations and rewordings: it is easier to reproduce. Hence the technology is a more robust memeplexe that will have less difficulties to survive and pass from mind to mind by copying.
Books
On The Origin Of Species - C. Darwin
The Extended Phenotype and The Selfish Gene - R. Dawkins
Wonderful Life - S. J. Gould
The Meme Machine - S. Blackmore
Internet articles
Viruses of the Mind - R. Dawkins
About memes - S. Blackmore
Meme replication and the memetic life-cycle - F. Heylighen
Amongst tasks devoted to the technical architect lies the choice of external solutions. This is a long process that involves the choice of suppliers, the creation of an evaluation grid, RFPs, work with suppliers relation teams, POCs and other evaluation prototypes, synthesis of all information and then presentation of all results to the team responsible for the final choice. During this process the architect and experts meet the suppliers to have demos and question sessions to dig into the internals of the proposed products.
One of the points of such an information gathering process is to ensure that the physical model for persistence is compliant with the context and constraints of the IS the product will be included in. This is important in physical layer but also in logical layer: are the concepts correct in regards of the IS needs? To answer this question the simplest way is to get the logical and physical models from the supplier. But they are never given. Instead questions about concepts are answered by showing how the GUI manages them. And so I find myself trying to rebuilt in my mind the logical model of the product the supplier presents to me. And all other architects and experts in the meeting are doing the same job. Don’t hope to end with the same model that way. Each time the debriefing session shows that we have understood things differently and that we don’t have assurances the concepts are really those we need.
Not to say that if they are not really what we’d hope there is nothing to do: the product forces its own concepts inside the IS.
Concepts are not what we need but what suppliers have decided we need. This is precisely the same behavior as proprietary formats for files: once we have stored our information in a provider solution built with closed formats, we’re tied. Because we don’t have rights to “reverse engineer” the formats (our mileage may vary from country to country, but still) we have to consider we’re no more owner of our data. This is the same for concepts: once we have bought the solution and the proposed concepts, we are tied to them. No matter what good analysis we may do, it will be very difficult if possible at all to include the results in the product.
It is also done by internal development teams when they are implicitly left with the semantic and logical models to do themselves. They capture concepts in what they have understood from the business and subsequent evolutions are difficult or even impossible.
By analyzing sooner the semantic and logical needs, Praxeme ensures that concepts implemented are those needed by the business. Obviously this works great for internal projects and this is the core usage of Praxeme. But it may also help software suppliers: they may either (or both :-) use Praxeme to create there products and they may create products that can be included in a composite IS while respecting Praxeme. This would be great.
Pre-built models (Currently a Praxeme initiative from Pierre Bonnet and Dominique Vauquier) are part of the answer: by using and participating in pre-built models internal teams and software suppliers are able to ensure that the concepts they include in there software are the good ones and offer the richness needed to build business centric IS.
On March 2007, Tristan Nitot has written a word about Praxeme. Even if it’s a little word and an old news, it is important because Tristan is a great defender of open standards and Open Source community.
In 1999 when I joined an Open Source community to port a software on VMS systems, this essay was a must reading for all Open Source fans. It represented some kind of a first description of methods used in Open Source groups. At this time Open Source was seen as parallel system for software editing with no relation to business one (that have been since proved to be false).
There are a lot of “principles” in this essay, like:
Some of them may be extended from the software development domain to the methodological one.
When I re-read it some days ago I was pleased to find this one that I forgot: “Smart data structures and dumb code works a lot better than the other way around“. This echoes my personal thoughts about what are more invariant in the real world we try to automate: things or actions?
Finally I recommend the page called “How Many Eyeballs Tame Complexity“. This could be a thinking direction for Praxeme further work.
This is one of my favorite movies.
Not for the poetry, I’m not very found of it. But for the main message: open-mindedness. When Prof. Keating climbs on the desk to explain how things that we know must be challenged by looking them from another point of view, I always have a special feeling of profound agreement. This idea is one of the reasons I on-boarded in the Praxeme initiative:
And then there is this scene where three students start walking at there own rhythm and end in the precise same tempo with all other spectators applauding with the same frequency. And Keating (interpreted by Robin Williams) shows how we are all different and why we should assume our differences. The idea is not new, but the scene is funny and the message is rendered very clear.
Not to say that the scenario is very good, pictures and lights are well done and actors despite their youth are doing a great job.
Since a long time (does it come from the beginning?) IT approach is centered on functions, those classical beasts that are considered essential when analyzing the needs for our systems. It is so important that the term does not only indicate an abstraction used to model the real but it is the retained term for denoting the business we are trying (and sometimes succeeding) to help. The “functional details” or the “functional view” relates to the business exactly as if we were saying “business details” or “business view”.
This semantic bias has happened without notice, as a slow migration from one sense to another very different. Semantic changes happen all the time, this is the nature of evolving languages. Here the speed and the degree of the change may seem astonishing.
Function is tied to action. When I “use” or “fulfill” a function, I do something, I act, I make a change in the real. No matter what this change may be, because I’m using a function, and this is good. Being? The thing on which the function acts? The thing I’m changing by acting? What’s that?
Seems a strange attitude? It’s not.
Because we as humans are here, now. And to be here, we need our ancestors to have passed a huge amount of obstacles by adapting to a huge number of situations where certain characteristics made them more resistant to adversity. Genes have passed those characteristics until us, modern human beings. And guess what: when we were only humanoids, or even some kind of monkeys, without notepads nor cellphones, when we were to meet a lion, what was the best according to Darwinism, run away or stay here considering what *exactly* is a lion?
Well now we know: action and functions are natural to us because this is partly why we as humans still exist today.
This post is clearly inspired by Richard Dawkins books (The Selfish Gene for example).
Last post on dvau
