I recently spoke at a PMI Chapter in Ballantyne event, well, I say spoke at but we really played a game of planning poker. We divided the group up into groups of four (that is how many suits per deck). There were 5 groups, and I supplied the prompt from which all responses were generated.  As expected,the initial prompt produced a wide array of results, from the lowest duration card available, to the maximum and in this case that was the infinity symbol – meaning not estimate able.    This was true not only from table (team at one table) but also from table to table.  After three passes, we could see convergence in each team as assumptions were evoked and questions were answered.

Every time I have used this technique, either in game or real life, I have witnessed this converging result. The team asks questions, more information is uncovered along with a variety of approaches and nuances to the work at hand are brought to the fore.  That is not to suggest that all of the answers of the team converge on one common value.  However, in many instances we can see the preponderance of results either common or within one card way. There can be outliers, representations further than one or two cards away from the clustering.  In this event, there was one group where the entire team eventually selected the same value estimate.  Now, that is not to say that this particular team has settled in on exactly what it takes to do this mythical work, but it does suggest that perhaps they have a common understanding and view of the work, and that no one person in this group, or any of the other groups, has unduly influenced the outcome by strength of personality.

Planning poker is much like Wide Band Delphi, but includes the entire team rather than selected experts, and is much quicker in generating results as the experts are not responding to paper questionnaire with the results compiled by the scribe to which the results are sent.  Upon compilation of the various perspectives into one document, the experts make another attempt at estimating based upon the input from the other experts.   This is all done in anonymity, no expert knows the input from any other expert.

Convergence may not mean that we have found the magic number estimate, but it does mean a certain measure of objectivity, and restriction of undue influence by dint of personality or strength of will.