ICANN ICANN Email List Archives


<<< Chronological Index >>>    <<< Thread Index >>>

RE: [gnso-wpm-dt] Introduction to draft Work Prioritization model

  • To: "Liz Gasster" <liz.gasster@xxxxxxxxx>, <gnso-wpm-dt@xxxxxxxxx>
  • Subject: RE: [gnso-wpm-dt] Introduction to draft Work Prioritization model
  • From: "Gomes, Chuck" <cgomes@xxxxxxxxxxxx>
  • Date: Sat, 21 Nov 2009 21:26:04 -0500

Great work Liz and Ken.  In my opinion this is very constructive for the
prioritization task in front of us. I strongly suggest that we
specifically discuss this approach as at least one option for
consideration. If anyone wants to discuss a different approach as well,
I am fine with that but it would be helpful if it is described
adequately to be able to compare with the approach described below.  I
also expect that we can work together to make additional improvements to
this proposed methodology or any others we consider.  In that regard,
see the comments I inserted below that we can discuss further on our
call on Monday.


        From: owner-gnso-wpm-dt@xxxxxxxxx
[mailto:owner-gnso-wpm-dt@xxxxxxxxx] On Behalf Of Liz Gasster
        Sent: Friday, November 20, 2009 3:42 PM
        To: gnso-wpm-dt@xxxxxxxxx
        Subject: [gnso-wpm-dt] Introduction to draft Work Prioritization

        Work Prioritization Team:  


        As a way to help bring everyone to the same level on the GNSO
Work Prioritization project, I have attempted to consolidate various
emails and organize our latest thinking into a single document.
Again, this is a suggested draft starting place offered by staff and the
group is encouraged to modify it as you feel appropriate.  There are
three sections as follows:  


        1)      Recommended construct and methodology (see also attached

        2)      Draft definitions for two dimensions  

        3)      Procedural questions to be considered 


        1)      Recommended construct and methodology


        For this effort, Staff is envisioning a two dimensional matrix
or chart (X,Y) to help the GNSO Council graphically depict its work
prioritization.  This concept is based on having each discrete project
rated on two dimensions:  Value/Benefit (Y axis) and Difficulty/Cost (X
axis).   Section 2 below outlines the preliminary draft definitions for
each dimension (or axis), so we will concentrate in this section on what
the chart means, how it would be produced, and the rating/ranking
methodology including sample instructions.  


        Illustration:   The chart below shows 8 illustrative projects
(simply labeled ABC, DEF, GHI, etc.) plotted on two dimensions:
Value/Benefit (Y axis) and Difficulty/Cost (X axis). 
        [Gomes, Chuck] I have no idea whether you anticipated this or
not, but it seems to me that a project naming scheme of three (or at
most four) uppercase letters could be very useful in at least a couple
ways: 1) It is concise so that multiple projects could be mapped on the
matrix; 2) It seems like it might be possible to develop meaningful
project identifiers that everyone can readily recognize without
constantly going back to a legend (e.g., PED for PEDNR, IRTB for IRTP
part B PDP).


        In this sample depiction, Q1, Q2, Q3, and Q4 represent four
quadrants which are drawn at the midpoints of each axis (arbitrarily set
to 10). 
        [Gomes, Chuck]  Note that the axes values are not calibrated so
that the quadrants are divided at actual midpoints but that is an easy
fix. Also, I think that the starting point should be zero instead of 1.
First, that helps solve the midpoint problem but it also allows for a
project to be rated as having no value, which theoretically could be
some rater's opinion.  So I recommend that we consider changing the
values on each axis to range from 0 to 10 with the midpoint (quadrant
divider) at 5.


        Thinking about Value/Benefit versus Difficulty/Cost, Q1 includes
those projects that have the highest value and lowest cost;
        [Gomes, Chuck]  I am not sure we should ever consider cost
alone.  In most cases it is probably likely that cost and level of
difficulty may go hand in hand but that might not always be the case.
It seems possible that a project could be very expensive to do but not
very difficult because it requires high cost resources while at the same
time being very straightforward to perform.  From another perspective, a
project may not cost too much but it may not be practically doable.  So
I would suggest that we always pair cost and difficulty together while
at the same time understanding that one factor (cost or difficulty) may
predominant in different circumstances. A project that has both high
cost and high difficulty would probably be rated higher on that axis
than one that had high cost but low difficulty.


        whereas, Q4 would contain projects with the lowest value and
highest cost.   Project ABC, in this example, is ranked 3.25 on
Difficulty and 7.75 in Value; therefore, it is located squarely in Q1.
Conversely, project GHI, is rated 7.75 on Difficulty, but only 1.00 on
Value and is thereby placed in Q4.   





        How do the projects end up with these individual X, Y
coordinates that determine their placement on the chart?  


        There are several options for rating/ranking individual
projects.   We will look specifically at two alternatives below:  


        Rating Alternative A:  


        One option is to ask each Council member, individually and
separately, to rate/rank each project on both dimensions.   Even with
this alternative (and B following), there are different methods
possible, for example, (1) place a ranking from 1 to n for each project
under each column, or (2) use something a bit simpler, e.g. High,
Medium, and Low to rate each project relative to the others.   Since it
is arguably easier to rate each project as H, M, or L versus ranking
them discretely from 1 to n, we will illustrate the former approach
here.   Keep in mind that an ordinal ranking methodology would simply
substitute a number (from 1 to 8 in our example) instead of the letters
H, M, or L.  


        Directions:    Rate each project on a scale of HIGH, MEDIUM, or
LOW for each dimension (Value/Benefit, Difficulty/Cost), but keep in
mind that the rating should be relative to the other projects in the
set.   There are no fixed anchors for either dimension, so raters are
asked to group projects as LOW, MEDIUM, or HIGH compared to each other.
A HIGH ranking on Value simply means that this project is perceived to
provide significantly greater benefit than projects ranked as MEDIUM. 
        [Gomes, Chuck]  Whether we use high/medium/low or a numerical
rating scheme, I think it would be very helpful to develop some
guidelines to facilitate rating and to hopefully improve consistency
across raters.  The guidelines could identify factors that may be
considered.  For the value/benefit metric, factors could include things
like these: % of overall community benefited; seriousness of the problem
being solved; etc.  For the cost/difficulty metric, factors could
include things like these: length of time needed; availability of
resources; etc.  These are just initial ideas that need to be developed


        If there are 20+ raters, we could provide a simple blank matrix
and ask them to provide their individual scorings.   For example, assume
that the matrix below is one individual's ratings for all 8 illustrative






























        Once we have all results submitted (could be simple Word, Excel,
or even Text Email) from all individual raters, Staff would convert each
LOW to a Score of 1, MEDIUM = 5.5, and HIGH = 10 (see attached
spreadsheet, Rankings tab).    We would then average the rankings for
all raters and produce a chart as shown in the attached spreadsheet (see
Summary tab).   Note:  We only used 4 raters in the spreadsheet for
illustrative purposes, but it is trivial to extend to as many raters as
we decide to involve.  
        [Gomes, Chuck]   I personally think that more differentiation
would be helpful than H/M/L would provide and hence lean more favorably
toward a numerical system. This is even more true in my opinion if we
are asking raters to compare projects when doing their rating.  


        Rating Alternative B:  


        Instead of asking each Council member to rate/rank each project
individually, the Council could use a grouping technique (sometimes
referred to as "DELPHI").  For example, suppose we set up 4 teams based
upon existing Stakeholder Group structures as follows:   


                    Team1:           CSG  = 6

                    Team2:           NCSG = 6

                    Team3:           RySG (3) + RrSG (3)  = 6

                    Team4:           Others (NCA, Liaison) = 4-5
        [Gomes, Chuck] Note that the scoring values above strongly favor
the Noncontracted party house.  We would need to carefully develop such
a rating method to maintain the balance between the houses and that
would add a level of complexity. 


        Using this approach, we would have 4 small teams and we would
ask for a single CONSENSUS score sheet from each one (whether ordinally
ranked or rated H, M, or L). 
        [Gomes, Chuck] Ordinal ranking might not result in accurate
values because it forces comparisons among projects that might not be a
good representation of value or difficulty or both.  That would in turn
impact the placement of projects on the matrix.


        Then, we would average those results to produce the overall
chart (similar to the example in the spreadsheet).   We should make it
clear that we are discouraging teams from individually rating and
averaging their own results.   The benefit, from this modified DELPHI
approach, is that individuals (especially new Council members) can learn
from each other and develop, collectively, what they think the most
appropriate answer should be. 
        [Gomes, Chuck]  It seems to me that we could encourage
collaboration in ranking whether we use this approach or not.  In fact,
whichever way we go, we might want to hold 30 minute overview sessions
with Q&A for each project.  Raters would only need to participate if
they were not familiar with a project at their sole discretion.  


        The above methodologies are subject to further discussion.
Ultimately, the Council will need to decide: 


        1)      What work prioritization construct should be utilized
(we have suggested a simple two dimensional Risk/Cost vs. Value/Benefit
displayed in a four quadrant model)?   

        2)      How should it be executed, e.g. participation, consensus
ranking (Delphi), individual ratings averaged, etc.? 
        [Gomes, Chuck] I think we need to keep as simple as possible
while still maximizing the meaningfulness of the results.  For that
reason I initially favor rating alternative A over B but am open to
further debate.  


        2)      Draft definitions for the X, Y dimensions


        Staff proposes the following starting definitions for the axes
in this conceptual model.  


        X - Difficulty/Cost ... this dimension relates to perceptions of
complexity (e.g. technical), intricacy (e.g. many moving parts to
coordinate), lack of cohesion (e.g. many competing interests), and,
therefore, overall cost/time to develop a recommendation. 
        [Gomes, Chuck] Note that you have identified rating factors that
can be used to develop rating guidelines as I suggested above.  This is


        We could have - but chose not to - create a third axis to
indicate the amount of time required.  This adds complexity and we
decided that initially we would include the concept of time into the
definition for level of difficulty.
        [Gomes, Chuck] I suggest avoiding the complexity and instead
consider 'time required' as one of the factors in rating


        Y - Value/Benefit ... this dimension relates to perceptions of
benefit to ICANN and its stakeholders in terms of internet
growth/expansion, enhancing competitiveness, increasing
security/stability, and improving the user experience.  
        [Gomes, Chuck] These factors can be used to create rating


        Please feel free to word-smith the above descriptions...


        3)      Procedural questions to be considered


        Once the matrix is developed and all projects plotted, what
should the Council do with the results?   This is an important question
to answer BEFORE the rating/methodology are finalized and executed. 
        [Gomes, Chuck]  We may only be able to partially answer this
question in advance because the details of the rating results will
probably affect the answer.  We should though be able to develop a
process for the exercise.  


        The Council should discuss and decide questions such as:


        1)      How often should it be exercised and/or what event
triggers an analysis?
        [Gomes, Chuck] Here's my opinion: It will need to be executed
initially for all existing work ; it will need to be executed for each
new project that is considered before a final decision is made on a
project; it probably should be done at least annually for all projects
on the table at that time.   

        2)      What decisions or outcomes are expected from the
        [Gomes, Chuck] Here's a start for an answer: a) The methodology
should be usable for prioritizing all current tasks and for making
decisions on new projects as they come up; b) lower priority projects
may need to be put on hold for a defined period of time or slowed down;
c) additional volunteers may need to be recruited before some projects
can be started. 



        Please let me know if we can provide any additional assistance
prior to and during the upcoming conference call on Monday.  
        [Gomes, Chuck] Another question we need to answer is whether or
not the first application of whatever methodology we decide to use
should involve all existing projects or some subset of them.  To
minimize the difficult I suggest we narrow the scope by weeding out
projects that we think should proceed as is. 


        Thanks and regards,







PNG image

<<< Chronological Index >>>    <<< Thread Index >>>

Privacy Policy | Terms of Service | Cookies Policy