Very thought provoking and also very provocative too, as I'm sure you intended! Just a quick acknowledgement at this stage. IT departments are in a strong position re Big Data because not only do they own and manage the data but they are also integrated into the management structure of their organisations, whereas OR groups are usually consultancy oriented and always seeking new projects/clients. In the UK, OR practice is very successful where it is established as witnessed by the Heads of OR Forum, the UK counterpart to the INFORMS Roundtable, consisting of around 40 heads of OR (and Analytics) groups. But many large organisations in the UK do not use OR overtly which is one reason for trying to capitalise on the current popularity of Analytics. It is not possible for a single Master's course to provide all of the skills required by OR and Analytics/Big Data practitioners and so OR must concentrate on where it is most effective. By the way, although I am now retired most of my career was as an OR manager in industry in the UK.
------------------------------
John Ranyard
Lancaster University
Derbyshire
------------------------------
Original Message:
Sent: 05-19-2015 09:40
From: Mark L. Stone
Subject: Should We Re-Brand Operations Research?
Big data/analytics has rendered Operations Research obsolete.
With big enough data, methodological matters such as
1) assumptions of Normality and other distributional assumptions
2) improper probabilistic independence assumptions and premature use of expectation (interchanging expectation with a nonlinear function) at an intermediate stage of calculation, perhaps combined with improper independence assumption
3) confusing P(A|B) with P(B|A) (Hi, are there any lawyers/judges here, ha ha)
4) accounting for uncertainty
5) linearity assumptions
6) dynamic evolution
are all rendered moot, and
7) explicit optimization is rendered irrelevant
because with big enough data, the best answer is automatically obtained and optimal, with no need for those O.R. geeks or Math type people trying to turn practical business problems into "science projects". If we need a Math person, we'll evaluate them based on how much Python, R, NoSQL, Hadoop, and Ruby on Rails coding they have done, because nothing else matters (well, maybe except for Java). If any of (1) through (7) appear to still matter, it's only because we need even bigger data.
Long live O.R. It was fun while it lasted. But big data has rendered it obsolete. Here's to getting unstable and unreliable and maybe flat out wrong answers with really big data, instead of with small data.
P.S. Numerical mathematical software experts (Numerical Analysts) are no longer required, because when performing machine learning, and computing things such as variance via numerically unstable textbook method (as was used in MAHOOT!!!!) instead of by a numerically stable one-pass method, all numerical instability problems are magically resolved and rendered moot when applied to large enough data sets, preferably using single precision on GPUs as is now in vogue. Algorithms which are numerically unstable and inaccurate when applied in double precision to 1000 data points, automatically become stable and accurate when applied in single precision to data sets with a billion points.
------------------------------
Mark L. Stone
Got into O.R. at the tail end of the glory days ca. early '80s
------------------------------
Original Message:
Sent: 05-19-2015 07:08
From: John Ranyard
Subject: Should We Re-Brand Operations Research?
Dear Diego and Rahul
You have both made some interesting points but I would urge Diego to read the Capgemini report commissioned by INFORMS (which I presume is on the INFORMS website) before dismissing it's conclusions, one of which is that the term 'operations/operational research' is little understood outside the OR community. On the other hand 'Analytics' has quickly become more popular, both generally and in management publications. Few groups that responded to my survey of OR practice for IFORS used the term 'OR' to label their service (preferring 'decision support', 'business analytics', and a variety of others which were found more helpful to potential clients). Also, Masters courses in the UK which rebranded as 'OR and Business Analytics' have become more popular with students than just 'OR' or 'Management Science'.
I believe that OR could become more widely known and understood by association with the decision making end of Analytics. At the same time we could enable those practicing in this area but without an OR background to become more effective by utilising OR's successful methodology which has been developed over many decades. As Shakespeare said: "a rose by any other name would still smell as sweet". We just need to ensure that a change of name does not reduce the scope of OR!
------------------------------
John Ranyard
Lancaster University
Derbyshire
------------------------------
Original Message:
Sent: 05-16-2015 07:24
From: Diego Ruiz-Hernandez
Subject: Should We Re-Brand Operations Research?
Dear John and all the rest,
I understand your point. But I am still waiting for someone that can give me a detailed account of the difference between Analytics and the interactions in Decision Sciences between applied statistics, data analysis, management sciences, probability or OR.
Regarding your concern about the potential customers. I think that, at the end of the day, the OR practitioner should provide a solution to his client, independently of the name that you give to it. Likewise, the firm manager looking for an answer to his problems cares more about the quality and price of such answer than about the name that you put on the technique used to find it. An argument htat is far away from the concerns of the Or theoretician.
OR is a science, or a branch of the mathematical sciences -as you prefer to call it-, not a collection of techniques waiting for the highest bidder. Analytics' emphasys on the application of such techniques to the solution of specific problems limits the wider scope of Operational Research.
Some time ago a university colleague made a clear statement -for her students- of the differences between OR and Logistics. And I keep thinking that her arguments -if I could only remember them- could be quite usefull to solve the confusion between OR and Analytics.
Analytics, for me yet to be proven not to be only a fashion does not matter what Capgemini says- is a 'sexy' given name for what has already been done by decision analysts over years: mixing OR techniques, statistics, data analysys and so on. On the other hand, OR is a discipline on its own, and the work of hundres of great men has bring it to a status in the mathematical sciences that can no be simply ignored.
Have a nice weekend!
------------------------------
Diego Ruiz-Hernandez
Associate Professor
University College for Financial Studies (CUNEF)
Madrid
------------------------------
Original Message:
Sent: 05-12-2015 07:19
From: John Ranyard
Subject: Should We Re-Brand Operations Research?
Hi Diego My interest is in OR practice and you need to know that detailed investigations by Capgemini (a consultancy that covers the full range of analytics) in the USA and UK, concluded that analytics is not just a new fad or fashion and that it had already achieved greater understanding and interest generally than OR/MS. Sadly only those within the OR/MS community understand and appreciate the full potential of OR. I believe that OR can benefit by association with analytics, but specifically 'decision analytics'.
------------------------------
John Ranyard
Lancaster University
Derbyshire
------------------------------
Original Message:
Sent: 05-11-2015 09:46
From: Diego Ruiz-Hernandez
Subject: Should We Re-Brand Operations Research?
Dear all,
OR is a more than 85 year old discipline. Analytics has not yet shown that it is something more that a fashion or a fancy word describing a collection of interactig disciplines.
I will be affraid of losing our identity. OR has been, and still is much more than Analytics. And yet, keeping our identity does not meen that we are not involved in or that OR is not at the heart of Analytics.
But that's just my opinion.
------------------------------
Diego Ruiz-Hernandez
Associate Professor
University College for Financial Studies (CUNEF)
Madrid
------------------------------
Original Message:
Sent: 05-10-2015 08:07
From: Rahul Saxena
Subject: Should We Re-Brand Operations Research?
Great post and interesting responses, both on informs.org and on the blog site. I see strong customer interest in using "analytics" to make better decisions, and they are open to the use of OR & DA techniques for this purpose. Like Thomas Groleau & John Ranyard, I think that an OR branding may not have brought me to these customers.
Removing OR branding in the face of the analytics-without-OR wave, on the other hand, risks re-tasking of OR practitioners into non-OR roles instead of playing the OR role in analytics-with-OR projects. So we may say that our options are (1) clearly separate OR from Analytics or (2) clearly integrate OR with analytics. In both cases we would drive an analytics-with-OR world, and separate ourselves from analytics-without-OR.
The issues that the http://www.humancenteredor.com/ blog raises possibly come from a sense that we're actually on option #3, that drowns OR into an analytics-without-OR sea.
In any case, can we drive a brand named "Operations Research and Management Science and Decision Analysis" or "OR/Decision Analytics"? They seem clunky. I think it may not be feasible, now, to lay claim to the "Business Analytics" movement that is merrily under way in an analytics-without-OR path. I hope we can come up with a nicer brand.
I'm using a "blue ocean" strategy in my decks to show that analytics-as-usual lacks the process to connect insights to decisions and decisions to results that we provide in an analytics-with-OR process. The OR/MS/DA community can brand the process but first we need to get our sub-disciplines aligned to build such a process. Branding requires identity, and it seems that we're struggling for ours.
------------------------------
Rahul Saxena
Cobot Systems
Bangalore
------------------------------
Original Message:
Sent: 05-07-2015 00:02
From: Robert Rose
Subject: Should We Re-Brand Operations Research?
http://www.humancenteredor.com
I consider wether we should re-brand operations research, and call ourselves analytics professionals.
Should We Re-Brand Operations Research?
------------------------------
Robert Rose
President
Optimal Decisions, LLC
Franklin Park NJ
------------------------------