Sunday, 30 September 2012

I am not an architect!

In the September issue of "Inside O.R.", the president of the Operational Research Society (UK), Geoff Royston, wrote about the image of O.R.  Inside O.R. is the monthly newsletter of the O R Society
Many years ago I wrote a short piece for Inside O.R. (back when it was called the O.R. Newsletter) in which I discussed the problem that O.R., as seen by others, suffered from some misclassification problems.
In particular that others tended to assume ‘operational’ meant tactical, that ‘research’ implied impractical, and that they too often equated the tools of O.R. (especially mathematics and computing) with O.R. itself.

I suggested that one approach to overcome or at least mitigate this problem would be to present our work so it aligned more clearly with a ‘general problem cycle’ that would be familiar to managers, The idea was that presentation of our work on these lines might provide a more marketable agenda for O.R. than one based on the tools of the trade or on narrowly technical problems.

To accompany that generic approach, I suggested that we might usefully present ourselves as ‘general analysts’. This drew on some analogies with general practitioners of medicine (who, in examination, diagnosis and prescription, follow their own version of the above problem cycle).

The analogy with medicine, particularly with the work of a G.P., seemed - still seems - useful. Firstly it places the focus on a client and their problem rather than on tools and techniques. Medicine needs and uses disciplines like biochemistry or physiology, but as a means to the practical end of helping a patient, not for knowledge for its own sake. Similarly, O.R. needs and uses disciplines like mathematics or computer science, not as ends but as means to its primary goal of helping with a client’s (or at least a potential client’s) problems.

Secondly, it emphasizes the importance of taking a systemic view. G.P.s need to take a holistic approach when considering the problems of, and interventions for, their patients, who frequently present with multiple conditions with more than one cause. A systems perspective is also essential in public health medicine, to help understand and control the spread of disease in a community. Just as the G.P. or the public health medic, by taking a broad systemic view, offers something distinct to patients and the public from those who specialise in a narrower area such as orthopaedics or cardiology, the O.R. professional, seen as a general analyst, offers something different to the world of management and organisation from that provided by those working in a single more specialised area such as economics or psychology.

So, medicine provides one useful analogy for us. But there is another profession with which analogy might usefully be drawn: architecture.

Like the O.R. worker, the architect is concerned with helping in a structured way (literally so in the latter’s case!) with a client’s problem, and also like the O.R. worker, the architect uses diagrams and models as virtual worlds on which to experiment. Like clinicians, architects follow their own version of the general problem cycle, but with a particular emphasis on innovation and the problems of design.

Design is challenging because it requires conceiving and assessing what can be a vast range of alternatives, with much iteration to find feasible and desirable solutions, because many design problems cannot be neatly subdivided into independent sub-problems, and because the process of designing, especially when done together with clients, often brings about changes in the perception of what needs to be designed.

Sounds familiar? The O.R. professional is often faced with problems that have something in common with the challenges faced by an architect. Although we may sometimes be asked simply to help with the choice between pre-existing options, more often we are given - or discover - a much more open-ended problem: the conception and creation of something that will improve a situation. That is a broad question of design; yet O.R. is persistently characterised as being primarily concerned with the important but narrower task of assisting decisions. This sells O.R. short.

The importance of design for operational research and management science was recognised in the 1960s by early luminaries such as the Nobel laureate Herbert Simon. He considered design to be the ‘core of all professional training’, including in management, and in his seminal book ‘The Sciences of the Artificial’ called for ‘ a science of design’ which would concern itself with topics such as the representation of design problems and the search for alternatives. He included in this science many of the then new methods of operational research.

Simon warned that design skills were being squeezed out of management training, and the later well-known series of papers on O.R. by Russ Ackoff noted that such skills were also being sidelined in management science. Ackoff argued that traditional techniques of analysis – techniques he had done much to promulgate - were insufficient for tackling important managerial and societal problems. He argued that also needed were the skills of synthesis His stated view was that the paradigm of O.R. should involve ‘designing a desirable future’ and his explicit challenge to O.R. professionals was ‘not so much to improve our methods of evaluation, but to improve our methods of design and invention’. Such concerns were noted and acted on by a number of people in the O.R community, at least on this side of the Atlantic. The most obvious result was the development of what came to be called ‘soft O.R.’. This sought to widen the role of the O.R. professional from technical expert with an emphasis on solving precise and sometimes unrealistic problems to that of reflective inquirer with an emphasis on framing and structuring problems of the real world in helpful ways.

This undoubtedly helped widen the vision of O.R. and did much to meet the criticisms of Ackoff and others, but the development of ‘soft O.R.’ appears more a response to the general criticism on the limitations of ‘hard’ technical approaches than to the specific call to accord a central role to design. There has been the occasional acknowledgement of the relevance of aspects of design thinking to O.R. – for example the strong but hidden similarities between engineering design and O.R. has been discussed by Albert Holzman, the implications for MS/OR of the shift in managerial interest from analysis for marginal improvements to design of complete systems has been noted by Robert O’Keefe and the relationships between operations management and design have been considered by Jan Holmstrom. Generally however such recognition has been sparse. A passable case for the defence, or at least a plea in mitigation, could be made that O.R. is deeply involved with design, but just does not make this sufficiently explicit. We should not forget that O.R. in the modern era was founded in a challenge of design - creating the British air defence information system that played such a crucial role in WW2. Design clearly continues to feature in O.R. work to this day - a look at, say, the winners over the years of the OR Society President’s Medal, or of the INFORMS Edelman award, shows that many of the projects have involved sizeable amounts of system design work. So maybe the main challenge is to recognise, publicise and develop this hidden expertise.

How might we market our design skills more explicitly and effectively? (The slogan developed to help publicise O.R., ‘the science of better’, hints at something that goes beyond decision making but the associated publicity materials have not highlighted design thinking). More importantly perhaps, how can we improve our design skills - how do we educate and train O.R. people better for these more creative aspects of our work? Developing our capability to embed technical problem solving and decision analysis in a wider activity of reflective inquiry and innovative design must surely increase our range, relevance and impact.

In operational research we need the diagnostic expertise of a physician. We need the architect’s skills of design. We are – or should be - both medics and architects.

In the following issue, my letter to the editor was printed, with the heading "I am not an architect"
In September's issue, the president suggested two analogies for O.R., medicine and architecture, to which you added the third of engineering.   
While I am happy with the first and the third, I have some reservations about the second.  
The analogy that an O.R. scientist acts as a doctor is a helpful one; the doctor's aim is to solve the problem - in this case, the problem of sickness or pain, and therefore enters a cycle of collecting information, intervening and participating in a feedback loop with the client (patient).  It is especially helpful because the natural association of the work of a doctor is to solve a problem.  There are limits to the analogy, particularly when one thinks of hospital specialists whose expertise is concerned with particular problems. 
Again, engineers solve problems, and they too enter a feedback loop of collecting information and data, modifying their design or product, making tests and evaluations with the aim of making something work successfully.  The downside of this analogy is that the first association one may make with the profession of "engineer" is not as a problem-solver, but as the motor mechanic or building site worker.  
There are many positive aspects to thinking about O.R. scientists and architects as coming from similar moulds.  Architects use models to think with and their models or diagrams may go through several iterations as they engage with other people; just like O.R. people.  Architects are concerned with a large system, and see the place of their work in an overall system; just like  O.R. people.  Architects are concerned with multiple academic disciplines and using different kinds of resources well; just like O.R. people.  Architecture combines art and science, a reminder for all of us that operational research involves creativity.  But, for many people, these are not the associations that will spring to mind when the job-title "architect" is mentioned.  That's where I have my problems with the analogy.   They are two, interlinked problems.  One is concerned about the definition of the client; the other is concerned with the timescale of the feedback to the architect.  An architect may have many "clients" for a project; the builder, the owner of the building, the people who occupy the building, visitors to the building, those who maintain the building.  A good architect will think of each of these, but will not be able to interact directly or indirectly with them all.  And there is no easy mechanism to alter the architect's work with feedback once the construction is complete.  Part of the best O.R. is that iterative loop of feedback and change once the clients (seen and unseen) have started to implement the results of the  O.R. work.  The second problem is timescale.  Generally an architect's work is there for many years.  Feedback is not possible for the users, owners, occupiers, etc, after many years.  (For example, my house is almost 80 years old.  I am unable to feed comments and suggestions to the architect about aspects of poor design of the house.)  There may be some O.R. results which last for years, but they are very few.  Perhaps a better analogy would be the designer of a fitted kitchen, something which is semi-permanent within the building. 
Hence, I am happy to be thought of as a doctor, engineer - but I am not an architect! 

Tuesday, 11 September 2012

OR54 - accommodation

Part of every conference-goer's experience is the accommodation.  In the UK, most O.R. conferences use universities as their venue, and so the accommodation is normally in student buildings.  Elsewhere, universities and conference centres are used as the venues, but many of the attendees stay in hotels.  (In the UK, the cost of conference-provided accommodation is sometimes more than the going rate for local hotels, so a significant number of delegates choose to stay away from the student accommodation.  Those who bring their partners/spouses will do this, as university blocks are full of single rooms.) 

I sometimes wish that I had kept a diary of my experiences when attending conferences; time has mellowed the memories.  There was the hotel in Beijing where my toilet would not flush -- and the staff member on duty on that floor was a young lady who did not speak English.  Somehow I indicated that I had a problem which needed to be dealt with, and she came to my room -- a middle-aged man with a young lady in a bathroom together!  And then there was the hotel in the red light area of Brussels ....

OR54 had very good accommodation - though it was pricey.  My two surprises were both in the bathroom.  First, the shower had no soap-tray, so nowhere to rest soap, shower gel and so forth.  Surely the designer knew that a shelf or dish was an essential in a shower!  Second, the toilet flush was so unusual that it needed instructions to tell you how to use it.  I think there was a good reason for that - the unusual design was a consequence of needing to save water.  But the instructions were only in English ... and OR54 had delegates from 20 countries.

Saturday, 8 September 2012

OR54 - the pub quiz

It is a tradition of O.R. Society conferences that there is a pub quiz one evening.  This happened in Edinburgh on the first day of the conference and nearly a hundred people took part.  It is a pleasant way to spend the evening with friends.  At the end of the evening (when our team didn't win, but didn't come last either) I listened to several serious or semi-serious suggestions for the quiz night in OR55.  So:
  • Try and make the questions accessible to an international gathering; there were at least 20 nationalities at OR54, and asking questions about literature meant that they were left out.
  • Make sure that the bar has plenty of draught beer - the bar ran out part way through the evening!
  • Have separate tables for each team - two teams to a table may lead to competition, but it is distracting.
  • Don't take the competition too seriously.
  • Have prizes which can be shared with the losers - boxes of chocolates and sweets are ideal
At OR33 we ran an alternative event to the pub quiz when I booked a harpsichord maker and performer to give a recital and talk about his craft.

Poor use of statistics in Edinburgh

While I was at OR54 in Edinburgh, I spotted the sign below:

Make our city Edenburgh.  Over a third of your bin is made up of food waste

Make our city Edenburgh.  Over a third of your bin is made up of food waste.

Obviously, the true statistic is that over one third of the total waste in Edinburgh is made up of food waste.  But that is not the same as pointing the finger at everyone and claiming that their bin is like that.   Maybe some PR person decided that this message would be more effective than something like:

Make our city Edenburgh.  Over a third of the city's waste is made up of food waste.

But neither message includes advice on what to do.  So, if I were an Edinburgh citizen, the sign would leave me annoyed (for the inaccuracy) and frustrated (for the lack of advice).

What do people who live in Edinburgh think?

Thursday, 6 September 2012

Lotteries are random

Of course they are - but however much you tell people that they are, some will doubt it.

Very little is published about the numbers that people have chosen for their lottery tickets, though in the UK (Choose six numbers from forty-nine) the most common combination is known to be 1,2,3,4,5,6.  When the Swiss lottery published some results over a decade ago, there were a lot of tickets which had chosen the previous set of winning numbers plus one - so last week's 5,20,26,29,31,39 would become 6,21,27,30,32,40.

There was mild interest in a rare lottery event in the UK this week.  In last Saturday's draw, the winning numbers were: 15,30,36,39,41,49 and five winners shared the jackpot prize.  Five is unusually high for the number of jackpot winners.  You expect a Poisson distribution, with mean about 0.7, which implies that if people choose randomly, there would be five lottery winners once in about 1000 draws.  But people do not all choose randomly, although I can't see what pattern has been chosen to give this set of numbers.  (There has been some research on how the layout of selection cards affects the choice of numbers - notably on the lottery draw in the UK when over 100 people shared the jackpot!)  This rare number of tickets sharing the jackpot was combined with the rare event of only one ticket winning the next prize, which in the UK is to match five of the six numbers and the so-called BONUS.  This is six times more likely than winning the jackpot.  However on 1st September, only one person won that prize, and as a result, that ticket won 50% more than the tickets that had shared the jackpot.  Coming second was better than coming first.

Is there any pattern to these numbers?

And a spokesperson for the National Lottery said "That's the beauty of the lottery - you never know what's going to happen" ... except that the organisers will make money from people who play, and from time to time there will be surprises.

OR54 in Edinburgh

I had seriously thought that my days of conference-going were over.  However, they are not.  A couple of months ago, I was approached and invited to be conference chair of the 2013 Operational Research Society (ORS from now on) Conference, OR55, to be held in Exeter in September next year.  Having done the job of chairing such a conference once before (OR33 in 1991 in Exeter) I knew this was an honour/responsibility/heavy load.  But I said yes.  Actually, I said yes to being the co-chair.  The ORS wondered if I would chair with a man named Philip Jones.  (cue for jokes about "Smith and Jones").  Who was this sucker volunteer?  It turns out that he is an Exeter graduate, who was on my courses umpteen years ago, and who was keen to work with me on the conference.  So you will meet his name again in the coming months.

One condition that I put on being co-chair was that I should be able to attend this years ORS Conference (OR54) in Edinburgh at no cost to me .... That will stop them, I thought.  No, the place was made available, so I went up on Monday, stayed two nights, and flew back on Wednesday because I had to speak in Plymouth in the evening on a long-standing booking.  The evenings and day in Edinburgh were full of meetings as well as conference sessions, with the aim of showing me what mistakes to avoid next year, so Phil and I can make a fresh set of mistakes at OR55.  (I am sure that every conference chair has their memories of mistakes ... or disasters ... but hopefully, not many other people remember them.)

I'll be writing about the conference in more detail over the next few days.  The venue was good, and well-equipped, the organisation ran smoothly.  I went to several sessions even though my retirement means that I was there without the excuse that the topic might be useful for my teaching or research.

I am pleased to report that the standard of presentation was generally high.  Everyone is using PowerPoint or similar for their illustrations, and the slides were generally put together well.  But, for future conference goers, here are some thoughts from the less than perfect shows.
(1) If you use an alarm clock on your laptop to remind you of diary dates, make sure that it does not go off during your presentation.  It will disrupt your show, and the audience will be distracted
(2) Don't put too much onto one slide.  PowerPoint is ideally geared up for about three points, and they each should be concise - a point is different from a sentence.
(3) Keep it simple, stupid
(4) There is a difference between the style used for presenting information in a written form, and in a spoken form.  One classification calls the text for the latter "Text that is written to be spoken" and includes sermons and public speeches (oratory).  The difference shows in many ways.  The lesson from this is that if you are going to read from a prepared text - perhaps because it will lead to a printed paper later - you should rehearse that reading with a critical audience.

And, to at least one delegate, I was introduced as the "OR blogger"

Indy's progress

Indy, the cuckoo about which I wrote about last month, has successfully crossed the Mediterranean, and was reported to be over the Sahel recently.  So the strategy of returning for food has been successful so far.