When I was applying for jobs, that is, faculty positions at universities, the search committees asked a lot of questions; about my research plans, teaching concepts, collaboration opportunities, about third party funding and my willingness to take on administrative duties. Besides this information gathering, I was also asked “more challenging” questions. One of them was
What were the greatest achievements in our field in the past ten years?
Our field, that is operations research, and in my particular case, discrete optimization and integer programming. Depending on the audience, my answer would have been more mathematical (“smoothed analysis of the simplex method“) or more business/applications (“revenue management”). The precise answer is not so important, but you should have one. Yet, good schools demand more of their leading scholars: they look for vision. So, some of them unavoidably asked
What will be the next big thing?
That is a tough one! It can be answered generally or personally. Depending on our angle, we would rephrase the question in different ways. Being a scientist, my interpretation is: Which major research question would you like to see settled next? Most research topics in my head deal with finding and exploiting structures in mathematical programs. I truely believe that we are not making best use of what modelers explicitly or implicitly encode in their integer programs. I expect from this to be able to solve much larger and more complicated models. But does this make it a big thing?
What are the grand challenges of operations research? My best way of answering this is to speak about opportunities, and the big loss if we miss them. If you look at emerging technologies (see Gartner’s 2014 hype cycle) there are several buzz words everyone is talking about, like analytics (yes!), big data, internet of things, fourth industrial revolution, etc. Non-specialists can connect to these notions, laymen have at least a fuzzy understanding of what these are all about, and: all consider them important. This is a mix that makes many companies jump on the bandwaggon, just to be part of it, they don’t want to miss anything.
But what is the substance behind these notions? What is the scientific foundation? Is this science?
I believe this is our field! Data is useless per se, even when big (maybe big data alone is even more useless than small data). We need methods to let data guide our best decisions—optimization methods. My fridge orders fresh milk, but according to which algorithms? I cannot speak of the #IoT without speaking about networks and network algorithms. Of what use is a digitalized production if I don’t know how to (best) operate the machines? Operations research and mathematical optimization offer models and methods to make significant and scientifically sound contributions here. Mathematics and algorithms are much needed to capture the buzz words’ true complexity and to answer questions like “how concretely should my internet of things function?” The answers won’t be always simple and this is the big risk: if we do not provide what we think are the right answers—operations research replies to the buzz words—someone else will. And believe me, someone else will have simpler answers, non-scientific answers, because these are much easier to digest and accept.
We should all work hard to establish operations research as the science behind big data, internet of things, industry 4.0, etc. We are the optimizers. We can make the best of these concepts. And we should.
Big thing!
café opt 
Interesting & motivational read!
Implementing the optimal solution and convincing the practioners to refrain from using the simple sub-optimal heuristics still remains an important issue in our field.
LikeLiked by 1 person
Thank you Sameer, I appreciate your comment. I sometimes say that we have a “marketing problem” — you can “see” our solutions only when implemented. But they won’t be implemented before you can “see” them. Vicious circle! So, let us all give examples, examples, examples of good OR work, so that finally many people “see” the benefits…
LikeLike
I have another suggestion: We need to consider operations and maintenance together – as a joint problem. Especially for large, complicated and expensive infrastructures (networks, resources, …), such as railways, roads, power distribution etc. Up til now, there has been a lot of focus on “operations” (after all we work with OR ;-)). Maintenance has gotten less attention, but mostly as a separate field from the planning and scheduling of operations.
Making better and more efficient use of scarce resources calls for methods that consider maintenance and operations together. These problems, I believe, are very interesting and provide a fine challenge for the community. Perhaps we should even talk about OMR or MOR?
(Yes, I know – maintenance can be viewed as an operation in itself, but that misses the point I think..)
LikeLike
Tomas, I am not sure whether this is relevant but if you google for “maintenance routing” you will find combined operations and maintenance decisions, for rail and airline applications. Maybe useful.
LikeLike
Yes, but these mostly concern maintenance on the carriers / vehicles / aircrafts (plus several concerning telecom networks). I rather think of the _infrastructure_ maintenance, which will temporarily degrade the service level (cancelling links etc). Similarly operations will usually hinder any maintenance activities. Thus both types of activities should ideally be coordinated / scheduled together. Especially since both account for large volumes and costs (in the previous mentioned infrastructure types). Tricky but important problems, I believe.
Sorry if drifting away from your subject, though. I fully support your opinion (thanks for a well written post!) – just wanted to highlight a less trendy field 😉
LikeLiked by 1 person
Marco, indeed a very interesting post. I fully agree that the areas – even though buzzwordish – of big data, analytics, and so on are right in the middle of our field, being the science of better decisions. But I think before we start giving ‘right’ answers we have to ask the ‘right’ questions. (A bit like ‘Game Theory’ being an answer to the question why a ‘self-organizing’ structure like the Internet exists.) What could these questions be?
Why is a ‘big input’ in any way different than an ‘ordinary input’? Why is the ‘Internet of Things’ different? Is it just marketing or something substantially different?
Being interested in mathematical questions, I think we need to start defining a model of computation that captures what the buzzworlds mean. In the case of ‘big data’ it could be something like: ‘if an unlimited number of processors is available how fast can you find a good solution?’,i.e., how much parallelism is there in the problem, or ‘if an algorithm is granted a certain amount of time, how much worse will its solution be compared to if it had ‘more’ time?’, i.e., something like a time-dependent approximation ratio.
So maybe we need a concept that restricts/enables computational resources in a different way than we were thinking about them before? I don’t know, but I am definitely interested … 🙂
LikeLiked by 1 person
Great comment, Alex. I believe that you express what I originally wanted to capture: the definition of, e.g., “the internet of things” should be done by those who are (at least potentially) able to methodologically deal with it. This is somehow similar to not letting people express “wishes” for a new software when they are not able to code. For big data also questions of redundancy could be interesting: does the “amount of information” (that is?) actually scale with the size of the data (and an advise could be: collect less data, or collect “good” data only, which triggers questions of “goodness”, of course ;-))
LikeLike
“We should all work hard to establish operations research as the science behind big data, internet of things, industry 4.0, etc. We are the optimizers. We can make the best of these concepts. And we should.”
AYE! 🙂
OR definitely have the tools to get value out of these new technologies.
LikeLiked by 1 person