In the January issue of New Scientist, there is a article about Parkinson's Law and how researchers in Austria put the law in a more scientific footing via mathematic. The essence of the law is, "work expands to fill the time available for its completion", which is intriguing but I am more interested in the second half of the article where Parkinson's limit is discussed.
Parkinson conjectured that there is a limit to any working group/committee size (20) beyond which no consensus would be reached no matter how the group is structured. I am curious whether this is also the limit of an agile/XP team size since an agile team tends to be of a flatter structurally (verses the traditional hierarchical nature).
Parkinson also noted that there is an anomaly around group with 8 members. He noticed that, for example, no nation has cabinet of 8 members. I have the good(?) fortune of working with teams sized on either size of 8 so I can't verify this claim. I wonder what are the proportion of previous failed or not so successful projects had exactly 8 team members. Then the question of how would one define team members. Would only developers, testers, BA count as a team member and not PM because they work very closely together day-in and day-out, whereas PM less so?
If you have previously (or currently) worked with team of 8 or larger than 20, what is your experience? Does the group dynamic change when the group size hit 8 or grew larger than 20?
Alex,Don't forget that Agile, by definition, has small teams (it's actually one of the agile limitations), so Parkinson's law probably doesn't apply. Maybe I will be challenged here, but give me an example of an Agile project where the team was or exceeded 20.Now I think Parkinson took into consideration everyone, including the PM when he made his calculation. The thing is the more team members you have the more communication channels you will have, and this thing goes exponentially. If you have 3 team members, then you will have 4 communication channels, if you have 4 then you have 9. I think the formula is m-1^2.In my opinion, a small team of 4 or 5 is ideal.
I agree on the 20 limit being unrealistic for agile team. But what about the size of 8? That is not so uncommon and I'd love to hear more about others' experience on that.
Moved to favorites. I'm gonna read it more often from now!
How do you define communication channels between team members? Are we
talking number of possible 1-to-1 communications which can occur in the
team?For a team with 3 persons the number of communication
channels would then be 3 (3 possible pairs in the team). For bigger
teams you can calculate the number of possible 1-to-1 communication
channels by calculation the number of possible pairs using the binomial
coefficient:(n!) / (k! * (n-k)!)where n is number of team members and k is number of members in a pair (always 2).(in Excel) =FACT(A1)/(2*FACT(A1-2))For a "team" of 2: 1 communication channelFor a team of 3: 3 channelsFor a team of 4: 6 channelsFor a team of 5: 10 channelsFor a team of 6: 15 channelsFor a team of 8: 28 channelsFor a team of 20: 190 channelsThis
is not an exponential growth, but a cubic growth. Please correct me if
I got something wrong here. Maybe I misunderstood the definition of
communication channels in a team. My apologies, I hope this might still
My post was more for those interested in the maths. I am not saying that it is all that relevent to this discussion.By a "cubic growth" I meant it grows by added dimension and is still more managable than something growing exponentially. That was my point.Nothing important, but if I see a formula I have to understand it 😉
Hi Matt,My definition of communication is not 1-1, it is m-n (other people may define it differently), where m and n are both bigger than 0, so let's say you have 3 members, A, B, and C.A->BB->CA->CA->B->C (A, B, and C are communicating together).A team of 4 will be:A->BA->CA->DB->CB->DC->DA->B->CA->C->DB->C->DA->B->C->Dfor a total of 10 communication channels, which proves that my formula above is wrong. Your formula is correct, but only works if you take into consideration that communication is always 1-1.
Aha, I understand. you have to keep summarizing the possible numbers of pairs, trios, quartets, etc as the team is growingFor a "team" of 2: 1 communication channelFor a team of 3: 3+1 = 4 channelsFor a team of 4: 6+4+1 = 11 channels (you left out A->B->D ;)For a team of 5: 10+10+5+1 = 26 channelsetc.I must admit I like thinking around this. Even if the maths are not useful, the model is interesting. probably the numbers should be weighted though. the number of pairs of persons who must communicate in a team are more useful than how many functional quintets you need in your 7 person team :PCheers!
I think that Parkinson's Law is pretty applicable in certain cases, like in large corporate businesses and the like. There are definitely cases, however, when Parkinson's Law doesn't really apply – here is a cool article on why it isn't necessarily always true: http://www.mindreign.com/en/mindshare/Global-Economics/Less-is-More/sl35291137bp353cpp10pn1.html