Monday 6 July 2009


Tacit or explicit transfer - which works the best? Another great exercise


Following on the theme of KM exercises and games, I was running a Masterclass today on Knowledge Transfer, and we did a little exercise, with some interesting results. The exercise went as follows

Three teams were given an identical set of 10 Lego pieces (plus a baseboard). For one team, the pieces were already made up into a structure, which we said was the Lego answer to the iPhone - the LPhone (the original LPhone is in the centre of the picture above). This team stayed in the main room, with this original structure in front of them, but hidden from the others behind a table.

The second team stayed in the main room as well, with their construction hidden behind another table, and were able to converse with the first team (though neither team could see the others' construction). The first team were able to talk the second team through how to build the LPhone, through dialogue and tacit transfer. The second team could ask questions to clarify the instruction they received. In the picture above, you can see the first team's original LPhone in the center, and the second team's "tacit knowledge transfer" version on the left. They are identical.

The third team were in a separate room. They were not able to talk with either of the other teams. Knowledge was transferred to them, from the first team, in a series of written notes. Knowledge transfer was therefore entirely Explicit - entirely through captured, written knowledge. Not only did the third team take more than twice as long to build their LPhone, the result (on the right of the picture above) was significantly different from the original in many ways.

SO why did the tacit transfer result in a faster, more effective knowledge transfer? When we debriefed the exercise we identified two success factors. Firstly the two teams (teams one and two) very quickly negotiated a common language to describe the lego pieces, and the orientation of the pieces (left/right, up/down, landscape/portrait). Secondly, whenever there was an inconsistency or ambiguity in the instructions, this could be resolved through questioning and feedback.

And why did the explicit transfer result in a slower, less effective knowledge transfer? Firstly there was no way to resolve any inconsistencies and ambiguities, so the receiving team (team three) had to use their best judgment (and one ambiguity early on had a knock-on effect on the design). Secondly, the order of the steps was not clear, as the notes which were sent through from the first team were not numbered. Finally there was no agreed language concerning orientation, or concerning the description of the lego pieces, which led to some errors.

SO what did we conclude? We concluded that knowledge should not be transferred only in written form, if tacit transfer (such as Peer Assist) is possible, though every participant reported that for their company, the default expectation seemed to be to capture the knowledge, rather than transferring it through dialogue. Secondly we concluded that there needed to be very good quality control on any explicit captured knowledge, to resolve any errors or ambiguities which might result in difficulties when applying this knowledge later.

This was a simple exercise, but with some very graphic learning points. It is one I recommend.

18 comments:

Mika Latokartano said...

It's an interesting experiment, thanks for sharing it! It also prompted a few thoughts.

Firstly, it reminded me of Dave Snowden's principles for KM, one of which states "We always know more than we can say, and we always say more than we can write down."

Secondly, it would've been interesting to have a fourth team, which would've received the instructions through dialogue, but would only have been allowed to build the LPhone afterwards, without the benefit of dialogue with the first team during the actual construction. This would've demostrated what things are important in verbal, tacit knowledge transfer (the successful use of narratives), and how people remember the things they have heard. What if that team would've had to pass on what they heard onto another team, immediately afterwards? Or if only one team member would've been allowed to have the initial dialogue, hear the instructions, and would've had to pass that on to the rest of the team? Useful knowledge tends to survive, it's like a meme. Knowledge that is not beneficial, 'dies'.

Thirdly, I'm somewhat against making either/or decisions, as they tend to lead to dogmatism and adversarial positioning.

Knowledge isn't transferred using just one way of communicating, but it's an organic and complex process. Methods that are used in knowledge transfer vary. It's important to understand the dynamics and the situation where knowledge transfer is required, and apply methods that work best in a given situation.

Like the old adage goes, "If your only tool is a hammer, every problem looks like a nail."

Finally a short story.

A couple of weeks ago I had to assemble a two cabinets and two bedside tables we bought from IKEA. They naturally came with instructions, which were actually quite clear, and assembling the cabinet following those instructions was actually quite simple.

Had there been no instructions, but instead I would've had to call a telephone number, where an expert would've walked me through the assembly on the phone, I don't think I would've been able to assemble it.

In this case, written, explicit instructions, worked better. It's a similar situation to the experiment you described, only more complex.

What was extremely useful, though, was to have my wife do the assembly with me. While quite clear, some parts of the instructions required a moment of thought. We both read the instructions, and during the assembly, had continuous dialogue, and complemented each others' understanding of the instructions. If I understood something wrong, she corrected me, and vice versa.

We were also quickly able to develop an efficient process for the assembly, for subsequent pieces of furniture. Having assembled one cabinet, our combined knowledge of the order of the assembly and of which working methods were the most efficient, allowed us to assemble the second cabinet without even referring to the instructions.

For the next two pieces, two bedside tables, we referred to the instructions, but had enough experience between us to understand the basic logic of how these pieces were designed to be assembled so that we only had to spend minimal time looking at them. Also our working methods were by then quick and streamlined, and we found that we could work together efficiently and without errors, without talking.

We had 'perfected' the assembly process in a very short time.

In this case, tacit and explicit knowledge transfer were combined, and there was a shared learning experience which culminated in mastering the assembly process in a very short amount of time.

At the end, we were talking about how the instructions and the construction could be improved, to further simplify the process.

New knowledge emerged from the simple process.

Nick Milton said...

I am not advocating "either/or" and I apologise if I gave that impression. I am a support of "both/and" (see here for example - http://www.youtube.com/watch?v=Vn6Yt9ZOwts where I talk about combining Connect and Collect in a balanced KM approach) and I share your belief that a blended approach, applying the right mechanisms in the right situations, is the best.

However we do have to recognise the risks in pruley explicit transfer, and take care to guard against them. The afternoon of the masterclass focused on how to address those risks, and how to construct and structure explicit knowledge in such a way as to address the risks.

Also it was interesting how many of the people at the masterclass were trying to work ONLY with explicit knowledge (largely I suspect because their KM implementations had been technology-led). This was a surprise to me, and it was worth restating the value of tacit transfer approaches such as Peer Assist.

Your suggestions for extending the exercise are very interesting, and I may well include some of these next time. if I do, I will report back here

Thanks

Mika Latokartano said...

I should've actually worded my reply better, because I didn't think you were advocating either/or. The exercise served as a tool to illustrate certain concepts, not to establish 'The Right Way' - my apologies for that, Nick.

I fully agree that attempts at purely explicit knowledge transfer as in so-called 'Best Practises' databases, and also in general, have great risks in them, and very rarely achieve the objectives set to them.

In tacit knowledge transfer much more than simple knowledge or information transfer is taking place, through the social interactions and behavioral patterns of those involved. Social links are established, the value of which might not at the time be apparent, but may prove serendipitous.

We work in human systems, being humans, and knowledge only really exists when it's in active in such a system. A document of any kind has the potential to evolve into knowledge, but it must be used, it must be an object of interaction in a human system. There it will also evolve, and adapt to the environment in which it is being used, or where it exists, maintaining its usability.

In an explicit form, for example in a knowledge base, it's in stasis, dormant.

The longer it's out of the human interaction of the tacit processes, the less relevant, the less adapted it becomes to the environment where it one day may be used. This, I think, is also why so many explicit knowledge transfer programs fail. A document from a knowledge base or a best practises database, having been revived from its stasis and inserted into a human system (a business process of any kind), may not be adapted to the environment anymore and either 'die', i.e. be entirely inapplicable unusable, irrelevant, or only partly relevant. It may and probably will perturb the human system, wherever it's required, causing distractions, which may result in inefficient perfomance or poor results -- with an entirely opposite results what are often sought after with best practise databases, knowledge bases, and so on. I feel that knowledge requires human beings to live, to be usable, and applicable. It must be active.

Maxim Grouchevoi said...

Hi Nick.
It's a very interesting story, thank you!

At the moment I am doing a research about intercultural communication of implicit knowledge (in Capacity Building Programms) - and I am surprised by the results. I found out that stronger focus on implicit dimension of knowledge sharing creates new ways of intercultural understanding and problem-solving (in capacity building programms).

So I am completely with you.

Nick Milton said...

Maxim, when you have the results of your research, it would be very interesting to see them! Thanks

Nick Milton said...

Re "Imaginary Time"'s comment on "the longer explicit knowledge is out of the human interaction, the less adapted it becomes" is an intersting observation.

I heard a story yesterday which may illustrate this. In the Nuclear industry, there is a small gadget they have used in the past, and is refered to in their explicit knowledge base. That gadget has gone out of production, but the same name is now in use for something rather different, which does almost the opposite of the first gadget. If you take the explicit base at face value and look for a gadget with this name, you could get a nasty surprise.

Tim Wright said...

Yes well it bears out the idea that knowledge is a verb and not a noun.
What you refer to as explicit knowledge is in fact information and it has only limited application in simple and repetetive circumstances. Hence the futility of capture and codification approaches to "KM". I would concede that more dynamic self generating information sources (possibly using social media) may have more richness and a longer shelf life ( and I probably find common ground with Imaginary Time here) but it is still very constrained in its value.

Nick Milton said...

Sorry Tim, I am still not totally convinced.

Both through tacit conversation, and through exchange of notes, there was some transfer of capability. The tacit route was quicker and error free, the exchange of notes was slower and error prone (in fact I would say the third team got about half the blocks in the right place, so you could say the transmission was about 50% effective. If you look at the photo, some of the message got through. The model on the right is not completely wrong, its about half wrong).

By both routes, the same "something" was transferred, but to different degrees. You can't say "one way knowledge is transferred, the other way it isn't", because the third team were able to half-build an L-Phone, so they half-got the knowledge. The best you can say is "one way knowledge was transferred, the other way it was half transferred", largely because of the loss of signal that came when the knowledge was codified and then decodified, and the inability to fill in the gaps through questioning.

Anonymous said...

Hi Nick - fun exercise! Hope Apple don't sue! :O)

I'm not certain though, that it really illustrates the distinction between Tacit and Explicit Knowledge. To my mind, it's a great exercise to demonstrate of the impact of synchronous dialogue vs serial monologue (a list of instructions).

The design of the exercise meant that the dialogue was verbal, and the instructions were written - but equally, the instructions could have been relayed verbally, and the dialogue electronic. Is a verbal list of instructions tacit or explicit knowledge? I think that Polanyi would have defined it as explicit knowledge, verbally transmitted.

My guess is that the *highest* performance would have come from a team which was given a photograph of the L-phone, which must count as Explicit knowledge - even if it is worth a thousand words!

I think then that the exercise shows that the choice of communication medium and degree of complexity of the task (hence need to negotiate a shared language/clarify questions)determines the performance, rather than the tacit/explicit nature of the knowledge/information.

I think Bird Island does that better!

Tim Wright said...

Nick I think where we differ is on what constitutes knowledge. As I don’t believe it is a thing then attempts to transfer it are, in my view, inevitably doomed. If we see it as an activity then what we are seeking to do is make that activity as productive as possible. What constitutes productive can vary on context. I would not disagree that information resources can be useful in making the activity productive in some areas but only in relatively simple transactions. I would suggest that your exercise demonstrated that even in a simple transaction - i.e. creating a replica of something these “explicit” information resources are of limited use. Now if you believe, as I do, that a true knowledge based economy needs to be considered as one where the knowledge transactions are far more complex and more demanding in value add, so for example in innovation environments, they I would say effort in creating "explicit" the information sources are of almost negligible value in a KM context.

Still an interesting post though

Mika Latokartano said...

Agreed, it's turned into an interesting discussion.

I think it's good not to take these kinds of exercises too 'seriously' in the sense that they would be designed to provide definite answers to complex issues, or provide yes or no answers. They are simple and trivialised by design, and their value lies in providing context and inspiration for further discussion and more detailed analysis. They're a springboard, of sorts.

Generally speaking, the whole tacit/explicit debate reminds me of the 90's KM discussion, where the whole organic and social nature of KM was vigorously attempted to fit into mechanistic constraints and rigid and artificial constructs of all shapes and sizes. With results I'm sure we all remember.

David Gurteen tweeted a link to Stephen Billing's blog entry on Five Reasons Why It Makes No Sense to Establish Corporate Values at http://www.changingorganisations.com/2009/07/five-reasons-why-it-makes-no-sense-to-establish-corporate-values/

One could almost replace the word 'values' with 'knowledge' in those five points, and it would make sense.

With knowledge, were're dealing with something that almost defies definition. Merriam-Webster has a definition for knowledge, but even in that entry, it's described with words like "condition", "familiarity", "acquaintance with", "awareness", etc. Knowledge is inherently bound to the individual, who he/she is, her experiences, connections, personality, culture and so on.

Knowledge exists almost as in a 'superposition' in us, as a range of possible states, conditions, actions or courses of action, and it only gets it's definite form when it's 'measured', used. Therefore it may be futile to attempt to codify knowledge. It only manifests itself when applied, in a given situation.

What can be done is codify fragments, be those instructions, guidelines, anecdotes, artefacts, which can be helpful. But actual knowledge manifests itself only through activities, like Tim Wright also suggested above.

Of course, this is also largely semantics, philosophy, epistemology and cybernetics. Which is what I suppose knowledge is largely about anyway. And fantastically interesting and inspiring!

Mika Latokartano
(Twitter: ImaginaryTime)

Stephen Billing said...

Thanks for your comment Mika.

How would you apply my second point, that values are always in conflict. Are you saying that knowledge is always in conflict? If so, what knowledge is in conflict with what? I am saying that one desirable value is always in conflict with some other desirable values. For example, the value in an A&E ward of seeing every patient within 4 hours could be in conflict with the value of dealing with life threatening cases first (triage). How would you apply this to the knowledge scenario?
Cheers, Stephen

Mika Latokartano said...

The analogy isn't perhaps perfect, but I saw clear parallels in that list, which inspired me. I think your second point initially suggested the idea of thesis-antithesis-synthesis triad as applies to knowledge.

In any problem solving situation people bring to the process their personal knowledge and experience, which is unique. The courses of action suggested by the knowledge and experience of individuals often vary, and through dialogue consensus can be reached, and new knowledge can be created. Knowledge that wasn't best adapted to the situation may during the process evolve, adapt, or die. The result of the process isn't an 'item' of knowledge, but evolved, adapted, or emerged new knowledge that is unique to every individual involved. Fragments of useful information, and other codified artefacts which may prove useful, may also be created.

I felt your third point supported this hypothesis, saying: "If everyone has the same values (knowledge [ML]), then you have a homogenous organisation in which the agents are very similar.

Differing 'knowledge agents' in the system must compete for survival (usefulness), and there's a greater chance of new innovations, ideas, and knowledge to emerge from the process than from a process with very similar 'knowledge agents' (which essentially are people).

But these ideas and this thoughtplay was perhaps even more suggested by the second part in "Values are always in conflict and are negotiated in particular situations." In any situation involving several people, the knowledge required to achieve what's needed, is always 'negotiated' through dialogue, where people bring their individual knowledge to the process.

I suggest it's far more useful to focus on the social dynamics in human systems than on attempts to codify knowledge. I agree with Tim Wright, who said above: "I would not disagree that information resources can be useful in making the activity productive in some areas but only in relatively simple transactions."

Knowledge transfer happens naturally, efficiently, and adaptively, when we don't impose artificial constraints, or impose knowledge transfer practises which are based on making trivialised assumptions on the nature of knowledge (I'm reminded again of the polarised tacit/implicit knowledge debate of the 90s).

Maxim Grouchevoi said...

Great conversation here!

I think the 90s debate around tacit/implicit knowledge was very influenced by Nonaka and Takeuchi (SECI) and used a dichotomy like told/untold or written/unwritten (remember the iceberg metapher) -> with the intention to "uncover"/externalize the hidden potentials (which is nonsense!). Today this debate has a new quality (see f.e. "Ba"-concept): and it is about the context/various contexts.

The thing is: knowledge is not only "unique to every individual involved" but also to every context (or multiplicity of contexts) involved. Sharing this contextes means sharing the knowledge - it's a very important part of the knowledge transfer. And there are at least three dimensions of it: cognitive context (person) - relational context (partner/opponent) - environment context (f.e. room/organization etc.). We have to redefine the meaning of "implicit/tacit" and - as I think - move back again to Polanyis ideas of "tacit" (who was misunderstood or misinterpreted).

That's why I like the story of Nick. It's first of all about different contextes - not told/untold OR written/unwritten knowledge. This is a true meaning of the implicity!

Sorry for my bad english. :(

Nick Milton said...

I am not sure whether tacit/explicit is enough. There are more subdivisions than this. There is unconscious, there is conscious but untold, there is told, and there is written. And even the "consious but untold" could be split into "conscious but unconceptualised" and "conceptualised". And to be honest, few if any of these steps are binary - most are gradations.

But I fear we are in danger of overanalysing a very simple exercise! It was a very useful and graphic demonstration of the challenges of transferring even simple capability in purely written form, compared to allowing dialogue and questioning. As Chris says, it delivers nowhere near the insight and emotional impact of the Bird Island game, but can be done in 15 minutes, it isolates and explores one single aspect, and allows some great discussion afterwards. I look forward to trying it again, and including some of Mika's suggestions

Mika Latokartano said...

Agreed, Nick, there's a danger of overanalysing a simple exercise, but that simple exercise proved to be a springboard not only for your masterclass, but to this interesting exchange of ideas and discussion here in your blog!

I say it well achieved what I imagine it was set out to do, and perhaps even more.

I also largely agree with your first paragraph there, and it reminded me of Dave Snowden's Seven Principles of Knowledge Management (http://www.cognitive-edge.com/blogs/dave/2008/10/rendering_knowledge.php):

"* Knowledge can only be volunteered it cannot be conscripted.
* We only know what we know when we need to know it.
* In the context of real need few people will withhold their knowledge.
* Everything is fragmented.
* Tolerated failure imprints learning better than success
* The way we know things is not the way we report we know things.
* We always know more than we can say, and we will always say more than we can write down." -Dave Snowden

It has been an interesting discussion here, thanks for allowing it on your blog.

jonathan said...

I have just heard about the idea of "explicit transfer". Both are really great and plus points I guess. But if to choose, I will choose the tacit knowledge..

Anonymous said...

Nick, excellent one, and sorry for coming late to the discussion - @tebbo pointed me here only today.

This reminds me again about the quest for "what is knowledge". A clear understanding would have steered group one better in their efforts to transfer such.

I do like Plato's definition a lot, it's quite practical (using it myself for IT process systems): "How objects relate to other objects" - keeping in mind that objects can be virtual or conceptual as well.

That's what the IKEA instructions are good at, especially as they know (duh) that adding context as in time or what sequence things should happen is important.

With a clear understanding as to what the task is (what to transfer), group one could have related each and every piece (object) using the grids (buttons) in question with precision, even without reverting to drawings.

Blog Archive