If you’re feeling overwhelmed at the world’s new-found interest in and demand for statistics, IMS President Xiao-Li Meng has a call to arms:
“I teach statistics” used to be an effective response when I was too tired to chat with a stranger, except that once it excited a taxi driver: “I teach mathematics too!” (It was on my way to the 2010 JSM in Vancouver, and I was grateful that we didn’t drive by any Mobius strip mall. It turned out that driver enjoyed traveling in topological spaces while being sleepless in Seattle, but preferred traversing in the Euclidian space during the fund-less summer.) These good quiet days are behind us. Now the same response would almost surely invite trouble: “Oh, great—what’s your view on AI?” or “Ah, so can you tell me why the 2016 election predictions were so wrong?”
Most human beings love to be loved, even when they cannot love back. But as a profession that loves conditioning, we should have a particularly good understanding that unconditional love almost surely does not exist, or at least it does not last. We have worked hard and long to attract the public and other fields to statistics and statisticians. Now they are infatuated and maybe even enamored with us (if you need statistical evidence, contact the statistics department chair at University of Toronto or at Yale University or at …). But how are we responding?
My checking of our pulse suggests that our collective heart is not beating as fast as one in the thralls of passionate love. (I sincerely hope that you have had direct measurements of that speed, and with few measurement errors.) I am acutely aware of the danger of making anecdotes plural. But since I have anecdotes at almost every resolution level, I at least can defend my emphasis on “Case after case after case …”
At the individual level, I have lost count of the times that I had to swallow that “Yes” bubbling up inside of me when invited to yet another intellectual or pedagogical getaway, knowing how many getaways that I already could not get away from.
At the local level, my colleagues and I lamented for years about not being informed of various university initiatives and activities involving statistics. Now I am getting a full serving of the mantra, “Be careful of what you wish for”. Every time I hear “Xiao-Li, we really need some statisticians involved. Can you or someone from your department help?”, I wish I were a ventriloquist. I know the only responsible answer would be “No.” I am already sleepless in Seattle, in Chicago, in Vilnius, in Park City, in Hong Kong, in Dublin, or in any other place a stochastic encounter has successfully enticed me to visit. And I’d be making the most obnoxious assumption that my colleagues are sleeping more, and hence I could volunteer them without feeling guilty. But I really don’t want the “No” to come out of my mouth, or those of my colleagues. How would we react if a group had complained for years that we had ignored them, but when we finally approached them, their response was effectively “Don’t you see we are all busy?”
At the national level, I was invited to the second annual Data Science Leadership Summit (October 12–13, Park City), representing IMS. It was a great gathering of many leaders of data science, mostly from university data science centers/institutions/initiatives. Among about 60 attendees, there were only four people who self-identified as statisticians. Was this due to the organizers’ disciplinary bias? Possibly for the first Summit, whose report stated: “Future incarnations of this Summit should be sure to include as equal partners other foundational disciplines of data science, e.g., statistics.” The Summit organizers did try, repeatedly, this second time around to attract us, but we statisticians are just too busy. Indeed, if I hadn’t been able to back out of a workshop I had previously committed to, I’d not have made it either.
These anecdotes become more alarming when combined with some depressing statistics. At the annual CATS (Committee on Applied and Theoretical Statistics, of the National Academies of Sciences) meeting, at 2018 JSM in Vancouver (no evidence of topologists driving taxis this time), and at the most recent NSF (National Science Foundation of US) workshops on “Statistics at the Crossroads”, I had conversations with multiple NSF program directors. They all urged IMS to encourage its members to play much more active roles “to take advantage of these opportunities and reach out to CS and Math or explore new partnerships with the domain sciences.” The opportunities they referred to include those listed in the article by the division Director of Mathematical Sciences, Dr. Juan Meza, to whom I am grateful for his willingness to accept my invitation to write directly to IMS members. Their urges are evidence-based, because the participation rates of statisticians and probabilists at the NSF level, from applying for grants to providing NSF with feedback, are significantly lower than those of more action-oriented disciplines such as computer science. In some cases, our participation rate is practically zero—how can we defend ourselves with zero participation rate when someone argues that we are essentially self-marginalizing?
“What are you talking about? I am just as sleepless as you are —what on Earth more do you want me to do?” If you are enraged by my suggestion of self-marginalization, then please join me to brainstorm how we can turn our enragement into engagement, collectively. There are only three strategies to address the challenge of demand exceeding supply: (i) increase the supply; (ii) reduce the demand; and (iii) use the existing supply more effectively. IMS has good opportunities to adopt (i) and (ii). For one thing, being an internationally leading scholarly society makes us attractive to other societies that are seeking to expand their efforts in data science. During that Data Science Leadership Summit, I met with the new executive director of ACM (Association for Computing Machinery), the world’s largest educational and computing society. The sheer size of ACM makes IMS o(1) with respect to every numerical metric I can think of: ACM has about 100,000 members, over 50% of whom reside outside of US, over 860 professional and student chapters with students participating from 500 colleges and universities worldwide, with about 300 annual conferences, etc. Yet ACM wants IMS as an equal partner in the intellectual pursuit of building the foundations of data science precisely because of our international status in leading theoretical statistics and probability.
I am happy to report that an IMS task force on the partnership with ACM has been established, with a charge to explore joint conferences, publications, membership, etc. Even if we succeed only in engaging and recruiting 0.5% of ACM membership, it would increase our current membership and capacity by about 15%! Another IMS task force is looking into extending NSF effort on behalf of statistical PhD education in the United States to international programs. Such pursuits enhance IMS’s outreach and education channels and efforts, as guided by (i) and (ii)—the more people are equipped with statistical and probabilistic insights and toolkits, the more supply of our workforce and the fewer people there will be who need to rely on others to deal with statistical and probabilistic problems.
Regarding (iii), there are fruits hanging low, or high, or currently too unripe. Whereas each of us may feel overwhelmed, collectively we have significantly more supplies, as long as we make a habit of trusting “the unusual suspects” (until someone proves to be not trustworthy), a relatively low-hanging fruit. For example, among all the committee chairs and members I appointed as IMS president, over 1/3 had never served on any IMS committees or task force. All these “first timers” responded positively, and typically more swiftly than “the usual suspects.” This reminded me that “supply” is a relative concept. Surely the “first timers” may have a longer learning curve, but all of us had our first time, and we owe our success to the trust of the generations that preceded us. I, for one, benefited greatly from the trust of the founding editor-in-chief of Statistica Sinica, Professor George Tiao (Happy 85th, George!), who asked me to help screening many submissions during my first year as an assistant professor.
Among very unripe fruits is a long overdue reform of our incentive systems, including the university tenure system. The current systems are not conducive for building broader pipelines (see my first President’s Column) or engaging in time-consuming collaborative efforts. A more effective system should explicitly recognize that transdisciplinary collaborative research typically requires a more holistic set of talents and skills to succeed than does within-disciplinary research. Having served as a dean, I know all too well why it induces a great laugh when the answer to “How many deans does it take to change a light bulb?” is “Change? Did you just say change?” Change is hard—otherwise I Love You, You’re Perfect, Now Change would likely not to be the second-longest running Off Broadway musical.
But love fuels change. Love induces and demands passion, and passion is most effectively expressed by action. If we really want the world to sustain its love for our profession, we must work collectively and creatively to harvest this grand passion fruit: the reprioritization of our profession to equalize our value systems for influential scholarly pursuits and for impactful collaborative effort, and to maximally reward those who can do both well. IMS is uniquely positioned to lead this reform on the global stage, and I therefore invite each of you to contribute in whatever ways you can to this (collective) labor of love, by love, and for love.