Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Entangling artefacts

Matti Schneider
November 28, 2014

Entangling artefacts

The concepts that allow an agile practitioner to leverage her artefacts to support continuous improvement through Entanglement: Emergent Semiotics, Reflexive Mapping, and Friction Isomorphism.

Matti Schneider

November 28, 2014
Tweet

More Decks by Matti Schneider

Other Decks in Research

Transcript

  1. Matti Schneider @matti_sg #XPdays 1 E N TA N G

    L I N G A R T E FA C T S Presented on the 28th of November, 2014, in Heeze (NL). CC-BY-NC-SA You can quote, reuse and remix this presentation as long as you cite me, but you may not make money out of it, including from training classes. If you include parts of this work in yours, you must make your own work available to the public for reuse too. Also, it would be nice from you to let me know about it :) If you want to use this material otherwise, please contact me.
  2. 2 Thanks for joining. We’re here to explore three properties

    of artefacts that I believe can transform them into supports for continuous improvement rather than being passive information holders. Goals: - discuss & criticize: find flaws, limitations in applicability - generate guidelines Please note that this workshop is experimental. Question the ideas, question the process. Perhaps we’ll be more comfortable diving into only one of the properties. That’s fine. I don’t know how it will end, and it is perfectly possible we find that none of this works (it would be a bummer, though). If that makes you more comfortable, you can consider you’re helping academic research :) What do I mean? I just have to give you a little background.
  3. Matti Schneider @matti_sg 3 I’m this crazy guy who decided

    that working in a startup was not challenging enough, and added a masters degree over it. Actually, I realized that knowing how computers work was not enough to develop software. I also needed to know how humans work. So I studied social and cognitive anthropology and chose agility as my research object.
  4. 4 Here is one of the walls of my main

    studied terrain. Covered in artefacts. How many artefacts do you see?
  5. 5 It is normal that there were different replies. There

    are multiple interpretations because there’s no single definition. Original definition: object made by humans and imbued with a specific meaning. Here’s the simplification we can make during this session: an “interactive information radiator”, software or physical. We’ll focus on the physical ones for this session. You can thus say it’s a “board”. Important, as we’ll see: artefacts are often described as cognitive aids, but that’s not what they are. They are part of the cognition in the distributed cognitive system that is the production team.
  6. Interactive information radiator 5 It is normal that there were

    different replies. There are multiple interpretations because there’s no single definition. Original definition: object made by humans and imbued with a specific meaning. Here’s the simplification we can make during this session: an “interactive information radiator”, software or physical. We’ll focus on the physical ones for this session. You can thus say it’s a “board”. Important, as we’ll see: artefacts are often described as cognitive aids, but that’s not what they are. They are part of the cognition in the distributed cognitive system that is the production team.
  7. Interactive information radiator 6 LIST! List the artefacts you’re interacting

    with daily. Timebox: 2 min. If not in a team at the moment, take whichever context you want. Will probably be easier to stick to it throughout the session. How many do you have? Who doesn’t have some variant of a task board (scrum board, kanban board…)?
  8. Entanglement 7 Name comes from quantum entanglement. “If I change

    the process, then the artefact is updated. If I change the artefact, then the process is updated.”
  9. Entanglement Modifying a property of the production system ≡ Modifying

    at least one of its representations 7 Name comes from quantum entanglement. “If I change the process, then the artefact is updated. If I change the artefact, then the process is updated.”
  10. To-do Implem entation Review Deploy Validation Done 9 After a

    few iterations, the team updates the board to reflect its actual process: “current” is split in the phases a software increment actually goes through.
  11. To-do Analysis Implem entation Review Deploy Validation Done 10 Updates

    are made iteratively on the process. When the team realizes it changed its way of working, it updates the artefact.
  12. To-do Exploto typing Concept ion Implem entation Review Deploy Validation

    Done 11 La phase d’analyse est donc éclatée en deux phases. La première, l’explototypage, combine deux approches pour récolter le maximum d’informations permettant d’alimenter la conception. Ce néologisme permet d’exprimer l’usage combiné de l’exploration (enrichissement en largeur de l’espace des solutions possibles) et du prototypage (enrichissement en longueur de l’espace des solutions possibles). La seconde, la conception, consiste en une prise de décision collégiale au sein de l’équipe, sur la base des informations accumulées lors de l’explototypage.
  13. To-do Exploto typing Concept ion Implem entation Review Deploy Validation

    Done 12 DoD DoD Concrètement, cela signifie que les étapes de conception et de revue de code deviennent des conditions de validation des étapes d’explototypage et d’implémentation. Celles-ci ne sont considérées comme terminées que lorsque l’équipe s’est accordée sur un point fixe.
  14. To-do Exploto typing Implem entation Deploy Validation Done 13 L’idée

    était de limiter l’exploration au strict nécessaire pour aboutir à un plan d’implémentation. De la même manière, l’implémentation est limitée à ce qui est nécessaire pour passer la revue de code. Cette modification du tableau a effectivement eu l’impact voulu sur le système de production, puisqu’une nette fluidification a été remarquée, entraînant une accélération et une baisse de la frustration.
  15. Descriptive 14 C’est la combinaison de ces deux aspects qui

    peut libérer la puissance d’un artefact.
  16. Descriptive 15 Il est donc possible de représenter la relation

    entre les artefacts et le système de production éminemment complexe (individus, logiciels, connaissance accumulée…) qu’ils modélisent sous la forme d’une boucle de rétroaction.
  17. Descriptive 16 Dans un premier temps, un artefact peut être

    descriptif : il rend visible certaines caractéristiques du système.
  18. Descriptive 16 Prescriptive Dans un premier temps, un artefact peut

    être descriptif : il rend visible certaines caractéristiques du système.
  19. Descriptive Prescriptive 17 Dans un second temps, un artefact peut

    être prescriptif : il commande, ou du moins influence, certaines caractéristiques du système. Lorsque les deux aspects sont réunis pour un même couple artefact / système de production, la boucle de rétroaction est complète. Le système entre alors en résonance avec son modèle.
  20. Descriptif Prescriptif 18 Entanglement Dans l’état de résonance, toute modification

    du système impacte au moins un artefact, et toute modification d’un artefact modifie l’état du système.
  21. Entanglement Modifying a property of the production system ≡ Modifying

    at least one of its representations 19 Friction isomorphism Emergent semiotics Reflexive mapping We’re actually creating an autonomous retroaction loop. Huge leverage for continuous improvement: reflect on the artefacts as representations of the process, change the representation, process is updated. We’ll shortly make this more actionable, but we can take a few minutes if there is a reaction already. Any questions?
  22. 20 Friction isomorphism Emergent semiotics Reflexive mapping Properties of artefacts

    that I believe to foster entanglement. Let me tell you straight that I am not proud of these names. They do sound cool, but they are not immediately understandable. They are precise, because I just finished abstracting away from observations. We are here to see how we can now apply these concepts. For each of them, I’ll give a definition, an example, a counter-example and then you’ll try to apply it to one of your artefacts. Afterwards, you will present your findings to the group and we’ll analyse the property.
  23. Representations of properties and parts in one artefact ≡ Representations

    of properties and parts in another artefact 21 Emergent semiotics Friction isomorphism Reflexive mapping Semiotics: system of signs, rules of combinations, ways of inferring meaning. Example: roman alphabet. How is the letter “G” pronounced in your language? “Consider the graphical language you’re creating as a first-class member of your system.”, thus “Treat your artefacts as a whole.” Cognition happens in artefacts through transfer of representations (Hutchins). Our artifacts remind us that we don’t only share the same language, we’ve also built it together. The tangibility of this togetherness eases transferring information when we use our shared language.
  24. 22 Example: story map. We can use emergent semiotics to

    outsource a part of the cognition in our artefacts. Not cognition as in “modification of the information”, but cognition by constraining the representation of the information.
  25. 23 With consistent semiotics, stories can be taken from the

    story map to the task board: they represent the same moving part through the production system. This action exhibits: - efficiency because the representations are optimized for a daily usage; efficacy because the team has created shared meaning through emergent semiotics.
  26. Representations of properties and parts in one artefact ≡ Representations

    of properties and parts in another artefact 24 Emergent semiotics COLLECT! Friction isomorphism Reflexive mapping 2 mins collection. Discussion.
  27. Representations of properties and parts in one artefact ≡ Representations

    of properties and parts in another artefact 24 Emergent semiotics COLLECT! What signs do you use? Do they have a specific name? Who gave it to them? Who uses these names? Is there a legend? Are symbols consistent across artefacts? Would it work if it had been imposed? By who? Friction isomorphism Reflexive mapping 2 mins collection. Discussion.
  28. Reflexive mapping One property of the production system ≡ A

    property of one artefact 25 Friction isomorphism “We write things that we know can change: if they never changed, we would just have to learn them.” It is easier to reflect on your rules when they are readable through your artifacts. Yes, making stuff visible to reflect on it is Kanban. But we often stop at the task board. Yet, the main production process is not the only important optimizable part of your production system. Your artifacts make not only your rules tangible, they often also show traces of how these rules evolved. Illustration beyond kanban board: mapping of the rules of interaction within the team, “Guide board”.
  29. 26 Il s’agit d’un artefact de type « Reflection workshop

    output », tel que défini par Alistair Cockburn dans Crystal Clear. C’est-à-dire qu’il est mis à jour lors des moments où l’équipe « réfléchit aux moyens de devenir plus efficace », pour lui permettre de « régler et modifier son comportement en conséquence » (cf. principe agile). En Scrum, ça signifie « à la fin des rétrospectives ».
  30. 27 À la fin de chaque rétrospective, l’équipe crée un

    ticket « guide » pour chaque règle qu’elle décide d’adopter, pour renforcer ses points forts et réduire ses points négatifs. Cela signifie donc que dans cet artefact est d’abord prescriptif. Il cherche à définir les règles du système avant que de les décrire, à l’inverse du tableau des tâches.
  31. 27 À la fin de chaque rétrospective, l’équipe crée un

    ticket « guide » pour chaque règle qu’elle décide d’adopter, pour renforcer ses points forts et réduire ses points négatifs. Cela signifie donc que dans cet artefact est d’abord prescriptif. Il cherche à définir les règles du système avant que de les décrire, à l’inverse du tableau des tâches.
  32. 28 Les premiers guides ainsi créés sont triviaux. Ils répondent

    aux problématiques simples du début, quand l’équipe se définit. Mais on a bien vu que pour qu’il y ait résonance, pour que la boucle de rétroaction soit bouclée, un aspect descriptif est nécessaire.
  33. 29 Cet aspect descriptif est apporté plus tard. Les guides

    ne sont en effet pas absolus, et peuvent devenir obsolètes. À une telle occasion, le guide est « fermé ». Exemple : la durée des itérations est changée au sprint 34.
  34. 30 Il est également possible que l’équipe n’arrive pas à

    respecter une des règles qu’elle avait édictées lors d’une itération précédente. Cette difficulté est repérée lors d’une rétrospective, et l’artefact est mis à jour pour refléter l’état du système. Ici, par exemple, l’équipe a régulièrement eu du mal à délimiter le standup à 6 minutes pendant l’itération, alors qu’il s’agissait d’un de ses guides. Pour que l’artefact soit un modèle du système, le guide en question est mis en « observation », une ligne spécifique du board. Cet aspect descriptif est couplé à un effet prescriptif pour l’itération suivante : le guide est mis en évidence pour l’équipe. Si elle n’arrive à nouveau pas à le respecter, elle en discutera lors de la rétrospective et pourra par exemple décider de fermer ce guide, ou d’en créer de nouveaux pour s’aider à le respecter.
  35. 31 Bien entendu, il ne s’agit pas de modéliser pour

    le plaisir de modéliser. Nous avons pu mesurer la puissance de cet artefact à travers le nombre de débats et de faux pas évités. Vous avez déjà eu une discussion où l’un des locuteurs pense que le débat a déjà eu lieu ? Dans notre cas, pointer le guide correspondant a toujours suffi à stopper net la discussion, sans aucune frustration puisqu’il s’agit de rappeler un point sur lequel il y a déjà eu accord. De la même manière qu’un ticket représentant une story est un « reminder to have a discussion », un ticket représentant un guide est un « reminder that a discussion has been had ».
  36. Reflexive mapping One property of the production system ≡ A

    property of one artefact 32 Friction isomorphism
  37. Reflexive mapping One property of the production system ≡ A

    property of one artefact 33 DESIGN! Friction isomorphism Design an artefact that represents an aspect of your representation system, by ensuring one of its properties maps directly to one property of the system.
  38. Reflexive mapping One property of the production system ≡ A

    property of one artefact 33 DESIGN! What makes your process different from another team’s? How does this difference show in your artefacts? What makes one of your artefact different from another team’s? How does this difference show in your process? Friction isomorphism Design an artefact that represents an aspect of your representation system, by ensuring one of its properties maps directly to one property of the system.
  39. Friction isomorphism Friction to modify a property of an artefact

    ≡ Friction to modify the represented system property 34 “It should be as hard to change a property of the artifact relatively to its daily use, as to change the function of the production process that is modeled.” Example: kanban board: easy to take tickets from one column to the other, much harder to add columns. Counter- example: issue tracking software!
  40. Friction isomorphism Friction to modify a property of an artefact

    ≡ Friction to modify the represented system property DESIGN! 35
  41. Friction isomorphism Friction to modify a property of an artefact

    ≡ Friction to modify the represented system property DESIGN! 35 When was the last time you updated an artefact? Did you update it to reflect your process or to drive it? How long did it take to apply the change? To change the artifact? Why didn’t you update it earlier? What should not change in your process? How hard would it be to change in your artefacts?
  42. Thanks! Guillaume Duquesnay Frédéric Merizen Huy Canh Duong Nicolas Dupont

    Thomas De Bona Photos CC-BY-SA Matti Schneider More : mattischneider.fr/anthropologie/agile/thesis 38 References : - Manifeste Agile, 2001 - Grammaire des conduites à projet, Jean-Pierre Boutinet, 2010 - Crystal Clear, Alistair Cockburn, 2003 - The Double Diamond Design Process Model, Design Council of UK, 2005 - Toyota Production System, S. B. Gershwin, MIT OCW, 2010 - Objects, Bodies and Gods, Arnaud Halloy, 2013 - Ethnographie d’un daily standup, Matti Schneider, 2013 - Toyota production system and Kanban system, Y. Sugimori, K. Kusunoki, F. Cho & S. Uchikawa, Toyota Motor Co., 1977 - Proceedings of « Artifacts, practice and knowledge elaboration » seminary, LAPCOS - MSHS - University of Nice Sophia-Antipolis, 2014