The Mathematics of Agile Dysfunction

“Who questions much, shall learn much, and retain much.” ~ Francis Bacon

In the movie Inception, Leonardo DiCaprio is a thief who steals corporate secrets by enter the a victim’s dreams. The plot follows his attempts at the inverse task of implanting an idea into the mind of a CEO. The problem with the technique is that the dreams can become so real that he no longer knows he is dreaming. His character uses a spinning top to know if he’s still in a dream state, or back in reality. If the top keeps spinning, Leo’s character is still dreaming. If it falls, he is awake. The top represents what Yuval Noah Harari refers to as part of an objective reality. It is something you can touch or feel or prove through mathematics in the case of the top’s physics. What is interesting is how much of human achievement is based on our imagination or subjective realities. Harari posits that what has allowed us to scale our collaboration globally is our unique ability as humans to create inter-subjective realities. Set of beliefs that are taken to be truths across entire communities of believers. Beliefs that can survive the comings and goings of individuals, sustained over time as well as distance. Working with clients using scaled Agile frameworks I find the same dynamics in play. What happens when you try and inject new scaled Agile memes into the existing corporate body of a culture? Will the body reject them, graft to them or be poisoned by them?

“The remedy is worse than the disease.”

A colleague sent me an email recently expressing his concern over the approach a project team was taking to Agile estimation while implementing a scaled Agile framework. I had the same concern. We both knew their notion was raising the hairs on the back of our Agile necks but we couldn’t find a constructive way to explain it. The issue was a common scenario in my world at least. The project team was being contracted to follow payment terms based on their performance and completed points were picked as the basis of payment. Immediately the team went about trying to develop a pricing model which amounted to finding the “cost per point” exchange rate. The client had used past history of points per release as the basis for how many they thought the development teams should produce over fixed periods. The project team quickly went into analysis paralysis as they tried to create a new economy based on establishing how many points each team must produce per release to keep pace with the costs of running the project. Conflicting with this was the preexisting level of effort model and T-shirt sized guesses used to figure out how many features could be jammed in the bread box of a release, regardless of invoicing concerns.

I used some of Harari’s observations to help me understand what is going on. Harari refers to the Cognitive Revolution as the period starting 70,000 years ago that led to a series of developments in thinking that put Homo Sapiens ahead of our evolutionary competitors and the elements in general. He argues that first we developed the notion of ‘us’ and ‘them’. It allowed us to collaborate beyond small social groups to those we had developed trust with. That trust still had boundaries. Early civilizations in and around 3000 B.C. still had borders with barbarian ‘them’s waiting in the next valley. The first millennium B.C. saw three universal orders emerge that allowed Sapiens to scale collaboration so that within these orders everyone was an ‘us’. The first to appear was the economic order of money. The second was the political orders that gave rise to empires. The last to appear was universal religious orders that spread beyond isolated communities, for example Christianity, Islam and Buddhism.

His point in the case of a global economy is that Sapiens have been able to develop a high degree of trust in something that is based purely on imagination. Bartering is difficult to scale and so was replaced with money. The invention of coins or replacing goods with some proxy thing of value was not uncommon. The Lydian Empires fame rests on its claim to have been the first state to invent and to mint coinage of a guaranteed value, replacing the barter system little by little with a monetary one that allowed trade beyond its borders. It became rich doing so giving rise to sayings like “As rich as Croesus” referring to one of its later kings Croesus (c. 560-547 B.C.E.). Our global economy has gone beyond coins, banknotes and cowry shells but this order is trusted the world over. Try to explain to a chimp how you paid for your last Amazon purchase when they are wedded to the physical reality of bananas. In Harari’s words:

“Money is accordingly a system of mutual trust, and not just any system of trust: money is the most universal and efficient system of mutual trust ever devised.”

It shouldn’t be surprising then when the team started created new memes in their imaginations that an economic model would feel most familiar. It’s so embedded so as to be natural but it is an imaginary inter-subjective reality.

“If a man’s wit be wandering, let him study the mathematics.”

The need to fit the new billing model of points into a precise math of money was natural but flawed. It was easy for the team to forget the original purpose of points and wrestle with the smaller problem of convincing themselves that points could be given a specific monetary value. In doing so they were embedding themselves deeper inside an imaginary memeplex like Dicaprio’s character in Inception. With increasingly less objective truths to hold onto, how does a team know what they are doing is helpful or harmful to the project’s community? Good people buried inside a bad dream. This then is my problem with prefabricated frameworks for scaling Agile. Groups are trained to accept the model but not given the tools to remain attached to some reality outside of the model. To be able to remind themselves that Agile is but a means. However you are bringing about organizational change is only a rite of passage, moving towards that end. The framework becomes the goal. Folks start talking about aligning to the framework and judge themselves accordingly. They introduce maturity models to rank their progress. It’s like going inside the nightmare tent of some Circus of the Damned, moving from being a spectator who can escape to part of an act and finally admitting the tent has no exit. The dream becomes is the new reality and absolute truth that no one will argue against.

“If a man will begin with certainties, he shall end in doubts, but if he will content to begin with doubts, he shall end in certainties.”

How do we escape the dream world if we think we are in it? Harari suggests that in understanding the universe, Sapiens have had to continually:

  • Admit Ignorance – accept the things we think we know to be the truth may not be and may be proven wrong as we gain more knowledge.
  • The centrality of observation and mathematics – having admitted ignorance we can set out to find new knowledge by gathering observations and connecting them to form new theories using mathematical tools.
  • The acquisition of new powers: Modern science puts the theories to use to create new powers such as new technologies.

I like that he uses the more inclusive ‘connected’ avoiding the assumption that the application of mathematics is directed purely at comforting crisp causality over the reality that we might be a part of a complex adaptive system. Those relationships may be statistical and not algebraic.

“Science is but an image of the truth.”

For the fun of it, I pick at the team’s model not to disprove it but just to understand how reasonable folks can convince themselves of a target mindset that will in fact create as much chance of harm as of good. What I show below is not meant to be a mathematical proof but more an attempt at a qualitative model to try and connect what is going on, to the familiar dysfunctions of this aspect of their transformation. Something to keep the flaps of the tent propped open. It reflects the talk I observed as they worked through the mental process without any of them stopping to do any math. It’s the wrong math for the situation but no one picked up on that. A tool like math applied incorrectly can be a part of the next layer of the dream.

The first point they decided to ignore is to consider that points are not anything more than a guess. Part proxy for the ignorance of how much a feature will really cost. An amalgam of effort and complexity, good enough to get agreement of the coding clan to move forward but in no way a commitment or prediction. Points like estimates have a statistical distribution not a precise value. We give them some value to make the math of stuffing a release easier. We collectively accept this falsehood as an act of collaboration not the creation of an illusion. The team however had grafted the familiar global order of money. They grasped at points as if they were objects with value. In their scaled Agile model, value is something observed outside of the team and its delivery system. What does occur inside said system is that a development team understands that over some period they can move some number of points to done as they define both points and done. Points are a shared subjective reality that extends no further than a development team. You cannot use them as the basis for other teams. If you do then that monetary universal truth tends to set in. Their purpose changes to a tool for judging teams, just as money is used to assess value of similar services from different businesses. Different subjective realities combined to create an inescapable false reality. The tent flaps disappear and minds become closed to external insights.

“In order for the light to shine so brightly, the darkness must be present.”

On the basis of this false reality the oversimplified math adds credibility but no link to objectivity.

Velocity v = points (p) / sprint length (s measured say days)

Expressed with such simplicity the project team loses touch with the realization that velocity is a made up number divided by an arbitrary length of time. Some of the scaled Agile frameworks have suggested the velocity be “normalized” across teams. Seems simple and reasonable given the sort of thinking that is familiar with pricing of gas/petrol within a locale. It also starts the injection of that notion of an imperial global order which suggests that development teams should naturally be compliant to the new order. Velocity becomes as a target. Taking a break from chewing on Bacon’s thoughts for a mouth full of W. Edwards Deming:

“People with targets and jobs dependent upon meeting them will probably meet the targets – even if they have to destroy the enterprise to do it.”

Now in order to figure out how much the project teams needs to charge for each point they considered how much it took to run the project including the folks not directly generating points. I don’t call them overhead. On these large crazy projects there are always good people smoothing the flow and removing waste, doing the contracting and recruiting. All that is independent of what the embedded development community is doing. It is the cost of all those bums on seats regardless of their role. It is the same idea as burn rate.

To run the entire program for some fixed period of time in hours T:

Cost C = Σ(Daily Rate (RB) x T). RB is a rate for some bum.

Now I calculate the average cost per point. I can mix the above into a slurry of subjectivity by making the period I’m calculating the cost over, the same as a sprint length i.e. s = T.

CP = C/p = C/(v x s) = Σ(Daily Rate (R) x s) / (v x s) at the point in time the cost-point exchange rate is established.

The project has to make money Moola (M) and only get to invoice an amount (I) based on Completed Points (PC). Yes, still no sign of whether the work had any positive outcome for the eventual customers.

M = I – C where I = m x (PC x CP) where m is some mark up. So the magic Moola mantra is then:

M = m x (PC x CP) – Σ(Daily Rate (R) x s)

Now if you are checking my algebra and thinking this formula will be useful, check over your shoulder and make sure you can still see the sunlight through the tent flaps.

If you think it is a load of crap that should be replaced with statistics, then good but stay with me because it oddly shows how some common dysfunctions will occur if the team can convince themselves it is true. The damn scary thing is we have seen a couple in action already and also watched the folks in question think their conclusions are perfectly reasonable.

“There is a wisdom in this beyond the rules of physic: a man’s own observation what he finds good of and what he finds hurt of is the best physic to preserve health.”

Dysfunction 1: If they don’t complete points (PC) in a release, then they start to lose Moola (M) resulting in the incorrect response from an Agile point of view: they will make anything have a “point value”. Defects have points even though the client never wanted them. Partial work like designs will have value. I’m not saying documentation is bad. Only that more will be created as a strategy to get low risk points to bill. When things go really south likely trips to the toilet and meetings will be worth points.

Dysfunction 2: If velocity increases, they get more points completed, billed and therefore they make more money. If that happens within a team, from finding better ways to work, then great! In a monetary model in which points mutate into a tool to evaluate relative efficiency of teams, velocity gets gamed to avoid consequences. Just enough Moola is made. The Circus rolls on.

Dysfunction 3: If improvement does happen then the client would expect the same outcome when a factory reduces the cost of goods. The savings get passed on in a reduced point cost. Gains are cancelled by reduced billings. There is then no rush to have any focus on improvement as innovation becomes a zero sum game. Remember no observations of the effect of improved throughput to the end customer make it inside the tent. Value has been applied to the wrong thing.

Most of the first three I have seen before. The next couple are from just looking at the equation and playing with the direction of values. Maybe you can see more.

Dysfunction 4: As you start scaling you need to add folks who don’t contribute points to keep the back-office business of the Circus running. The cost of running the Circus goes up as more bum rates are summed. To keep up the Cost per point must inflate. The project team will then have to explain every few months that they are selling “New & Improved” points that are just more expensive versions of the same old points.

Dysfunction 5: If the development teams are continually not producing the completed points planned, then the project will need to start to get rid of people who don’t produce points. Compliance work like required documentation and managing everyday project noise now falls to the teams. This will cause further decay in points being completed until the system implodes.

The challenge faced by this team in estimation is merely a microcosm. A single act within their scaled Agile Circus of the Damned. It’s like the Scrum top they spin has sides numbered with Fibonacci series and in their dream they hope to see it fall on one numbered side every spin. The idea that it should land on a vertex and sway back and forth is too foreign to be accepted into their newly imagined scaled Agile world view. Ideally they would want the top to be one that spins endlessly and for them to be concerned only with improving the local environment so the top can spin faster without direct intervention, without wobbling.

Harari points out that when cultures appeared as part of the evolution of human society, they were not a stagnant target state. They never ceased to mutate. The timeline of the alterations and adaptations to a culture is what we call history. Agile mindsets, like any other imagined reality and the associated ceremonies and practices are just a locally emergent culture. There is no one Agile way. There is no “Being Agile”. Scaled Agile frameworks appeal to a notion of universal social order. A successful Agile transformation then aims to limit a potentially vibrant organization to a set of well recognized mass produced Agile antiquities. Better to let folks create their own culture and let it its history play out.

“Antiquities are history defaced, or some remnants of history which have casually escaped the shipwreck of time.” ~ Francis Bacon

Advertisements

About guywinterbotham

An Agile Buccaneer navigating the corporate storm
This entry was posted in Agile Anthropology and tagged , , , , , , , . Bookmark the permalink.

Leave a Message in this Bottle

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s