Slide 1

Professor Michael Mainelli, Executive Chairman, The Z/Yen Group

[An edited version of this article appeared as "Betting On Proper Procurement" Xoomworks Blog, Xoomworks (November 2012).]

Here’s a short tale in three acts spanning two decades, with one conclusion.

ACT I – Anna Karenina (if only it were a single act)

In October and November 1992 the London Ambulance Service exhibited a public ICT implementation catastrophe. My co-partner at Binder Hamlyn, Paul Williams directed and wrote the Report on the Inquiry Into the London Ambulance Service (LAS) (February 1993) and I became the quality assurance partner responsible for turning around LAS ICT over the following two years, though most of the inspirations for success came from one of our managing consultants, Chris Webb. For years afterwards I have seen consultants brandishing the report at events saying, “The LAS report shows that the single most important thing in new ICT procurement is…”, followed by a sales pitch for their respective service – an ICT methodology, project management, training, staff involvement, technology risk assessment, … They miss the point, or are mis-selling. A cursory reading of the report, and the causal diagrams we deliberately inserted, shows that many things went wrong at once. LAS was an exhibition of Jared Diamond’s Anna Karenina Principle, from the opening line of Tolstoy’s novel: “Happy families are all alike; every unhappy family is unhappy in its own way.” The Anna Karenina Principle describes situations where a number of activities must be done correctly in order to achieve success, while failure can come from a single, poorly performed activity. Bad procurement is subject to network effects, particularly when negative feedback is insufficient or slow, and positive feedback is too strong.

ACT II – Placing our bets

In 1999, at Z/Yen, a large music company asked us for an innovative way to teach young people about finance. We designed and built a six million user betting game on online music sales. Players were rewarded with dividends based on the number of downloads for ‘shares’ in the tunes they bought. If a tune had a high price/dividend ratio, players thought it would do well. If players thought a tune would do poorly, its price was low. When the internet boom burst in 2000, with a “the web is dead” (!) wave goodbye, they failed to launch the game. But we kept plugging away on predictive markets and run a number of them to this day on our ExtZy platform (free to play and with prizes – We use ExtZy games to assess sentiments in areas of interest to us, such as views on new technology or disproportionate country news. Predictive or betting markets have many uses, particularly when analysing network effects such as technology futures or news. During the 2000’s, I was approached by three enormous research and development organisations. In all three cases their problem was that their teams were unaware of work within their own organisations, leading them to ignore their own internal experts and contract expensively and more poorly outside the organisation. In all three cases we showed them how they could use a market that ‘bet’ on the popularity of their thousands of internal projects to make learning fun for their teams, provide inter-team connections for intranet tools (e.g. an app that helps you design a pill coating so you use internal specialists’ expertise), and help managers assess genuine sentiment about colleagues’ projects. Not that they did it.

ACT III – Buying versus shopping

Ian Harris and I have written on the importance of shopping before buying. Certainly set some procurement criteria but then go window shopping so criteria can evolve. Shopping is about learning and inserting appropriate diversity. Buying is about the transaction. But with network effects, procurement gets tricky, so perhaps we need a more robust approach. In the 2010’s, our firm, Z/Yen, has been approached for some innovative thinking on two very large, long-term, UK infrastructure projects. Both projects complained that they were being ‘gamed’ by suppliers, and were having numerous delivery problems centred on time, i.e. consistently late delivery. In both cases there were three large suppliers who were fulfilling the technical specifications to the letter, but the overall project was suffering. For example, contractors were asked to build structures at a particular location, and did. But then it turned out that the contractor knew all along that the structure was over a problem foundation, but failed to report it. The structure had to be abandoned and a new site found. Of course the contractor took the money and waited for the next request to build. Classic contracting. It got worse when things like running telecommunications sub-systems through multiple structures depended on all the structures being ready. Everybody was buying, but nobody was shopping.


On any project, there are four major dimensions – scope, quality, cost and time. Unrealistic timings put scope, quality and cost at great risk. In both cases we showed the infrastructure projects how they could build a cooperative betting market on time. Scope, quality and cost were covered by normal procurement methods, and BS 11000 on cooperative business relationships was helping somewhat. Our suggested approach to the time problem was to talk with suppliers about a sensible margin for scope, quality, cost and time, say 12%, then tell them 12% margins were assured. While 8% of the margin would be theirs on normal terms as work progressed, 4% of the margin had to be placed on the betting market and was available only at the end of the project. The bets would be on structure completion times, multiple structure completion times (e.g. a region) and on sub-system completion times. The bets would be ‘outcome’ bets that fluctuated between 0 and 100. Such a market would give the client an idea of what individual contractors truly thought about delivery times, and what contractors in aggregate thought about delivery times. Given the sums involved this much more than justified having one or two people from each contractor betting full-time. Such a betting market would involve about eight people full-time. In theoretical terms, the market would elicit other hidden information and reduce information asymmetry problems on a large-scale contract. People would be learning. Not that they did it.

One of Xoomworks’ clients had a procurement department that the organisation viewed as a “place where good ideas go to die”. It’s about time we in procurement learned that our increasingly complex world needs simultaneously more complex systems, such as a learning procurement environment, that evolve towards simpler interfaces, such as a betting market. As a species our progress relies on exploring complexity, then learning how to simplify it – air travel: learn how to fly, put on lots of instrumentation, experiment copiously and dangerously, then make it simpler and safer. Perhaps that time is coming soon to procurement. I’m betting on it.

Professor Michael Mainelli FCCA FCSI FBCS, Executive Chairman, Z/Yen Group

After a career as a research scientist and accountancy firm partner, Michael co-founded Z/Yen, the City of London’s leading commercial think-tank, to promote societal advance through better finance and technology. Michael’s third book, based on his Gresham College lecture series from 2005 to 2009 and co-authored with Ian Harris, "The Price of Fish: A New Approach to Wicked Economics and Better Decisions", won the 2012 Independent Publisher Book Awards Finance, Investment & Economics Gold Prize.