Like many other people who have been watching the COVID-19 press conferences held by Trump and Cuomo, I came away with a very different feeling from each. Beyond the obvious policy and partisan differences, I felt there is something more going on.
Coincidentally, I’ve been doing some research on text analytics/natural language processing on a different topic. So, I decided to use these same research tools on the transcripts of their press conferences from April 9 through April 16, 2020. (Thank you to the folks at Rev.com for making available these transcripts.)
One of the best approaches is known by its initials, LIWC, and was created some time ago by Pennebaker and colleagues to assess especially the psycho-social dimensions of texts. It’s worth noting that this assessment is based purely on the text – their words – and doesn’t include non-verbal communications, like body language.
While there were some unsurprising results to people familiar with both Trump and Cuomo, there are also some interesting nuances in the words they used.
Here are the most significant contrasts:
The most dramatic distinction between the two had to do with emotional tone. Trump’s words had almost twice the emotional content of Cuomo’s, including words like “nice”, although maybe the use of that word maybe should not be taken at face value.
Trump also spoke of rewards/benefits and money about 50% more often than Cuomo.
Trump emphasized allies and friends about twenty percent more often than Cuomo.
Cuomo used words that evoked health, anxiety/pain, home and family two to three times more often than Trump.
Cuomo asked more than twice as many questions, although some of these could be sort of rhetorical – like “what do you think?”
However, Trump was 50% more tentative in his declarations than Cuomo, whereas Cuomo had greater expressions of certainty than Trump.
While both men spoke about the present tense much more than the future, Cuomo’s use of the present was greater than Trump’s. On the other hand, Trump’s use of the future tense and the past tense was greater than Cuomo’s.
Trump used “we” a little more often than Cuomo and much more than he used “you”. Cuomo used “you” between two and three times more often than Trump. Trump’s use of “they” even surpassed his use of you.
Distinctions of this kind are never crystal clear, even with sophisticated text analytics and machine learning algorithms. The ambiguity of human speech is not just a problem for machines, but also for people communicating with each other.
But these comparisons from text analytics do provide some semantic evidence for the comments by non-partisan observers that Cuomo seems more in command. This may be because the features of his talks would seem to better fit the movie portrayal and the average American’s idea of leadership in a crisis – calm, compassionate, focused on the task at hand.
A virtual mirror allows someone to use a camera and have that image displayed on a large LED screen. Better yet, with the right software, it can change the image. With that ability, virtual mirrors have been used to see what new glasses look like or to try on dresses – a virtual, flexible fitting room.
Virtual mirrors and their equivalent as smart phone apps have been around for the last couple of years. There are examples from all over the world. Here are just a couple:
This all provides a nice experience for customers and may even help sell a particular item to them. But that’s only the beginning.
Virtual mirrors are a tremendous source of data about consumer behavior. Consider that the system can record every item the consumer looked at and then what she or he bought. Add to that the information about the person that can be detected – hair color, height, etc. With the application of the right analytics, a company can develop insights about how and why some products are successful – for example a particular kind of dress may be what short or tall women are really looking for.
With eye tracking devices, such as those from Tobii, connected to the virtual mirror, even more data can be collected on exactly what the consumer is looking at – for example, the last part of a dress that she looked at before deciding to buy or not to buy.
Going beyond that, an analysis can be done of facial (and body) expressions. I’ve written before about affective computing which is the technology is developing to do and to respond to this kind of measurement.
By fully gathering all the data surrounding a consumer’s use of the virtual mirror, its value becomes much more than merely improving the immediate customer experience. In a world of what many consider big data, this adds much more data for the analytics experts on the marketing and product teams to investigate.
Alas, I haven’t seen widespread adoption and merger of these technologies. But the first retailer to move forward this way will have a great competitive advantage. This is especially true for brick-and-mortar retailers who can observe and measure a wider range of consumer behavior than can their purely e-commerce competitors.
Among the more ambitious and expansive CEOs, there’s a special kind of holy grail – transforming their organizations into learning organizations. Jack Welch, former and famous CEO of GE, put it this way in the 1990s:
“an organization’s ability to learn, and translate that learning into action rapidly, is the ultimate competitive business advantage.”
“Organization that acquires knowledge and innovates fast enough to survive and thrive in a rapidly changing environment. Learning organizations (1) create a culture that encourages and supports continuous employee learning, critical thinking, and risk taking with new ideas, (2) allow mistakes, and value employee contributions, (3) learn from experience and experiment, and (4) disseminate the new knowledge throughout the organization for incorporation into day-to-day activities.”
Or as Peter Senge, one of the founders of the learning organization movement, has famously said in 1990:
“A learning organization is an organization that is continually expanding its capacity to create its future.”
As you can see, this dream started more than 25 years ago.
By the first decade of this century, however, Mark Smith wrote:
“[W]hile there has been a lot of talk about learning organizations it is very difficult to identify real-life examples.”
The companion in thought of the learning organization movement was the knowledge management movement. Its goal was to capture, organize and distribute knowledge among the members of an organization.
In some ways, this reminds me of the first great wave of Artificial Intelligence in the 1980s when a lot of effort went into trying to codify the knowledge of experts into expert systems by extensively questioning them about their decision processes. It turns out that it is hard to do that.
Often experts – the knowledgeable ones – can’t really articulate their decision rules and to make matters worse those rules are at times probabilistic. Like other humans, experts often seek to develop rubrics to simplify a problem, which unfortunately can limit their ability to continue to observe what’s happening. Human perception, in general, in an imperfect instrument.
Thus, even if an organization is successful in widely distributing the knowledge developed by its staff, it may be just propagating ideas that are, at best, only partly true.
All of these factors, and many others, has slowed down the march to the dream of learning organizations.
But now we are possibly at the beginning of a rebirth of the learning organization.
What’s different today? Analytics and big data make the process of organizational learning much easier and better. Indeed, it is perhaps time to add analytics as a sixth discipline to Senge’s Five Disciplines of a learning organization.
After all, it’s not that people don’t know things that are important to an organization’s success – it’s just that they don’t know everything that is important and they can’t or won’t keep up with the torrent of new data vying for their attention.
The traditional gathering of the human knowledge combined with the continuously improving analytics models can achieve the dream so nicely stated by the executives and visionaries of twenty-five years ago.
For example, instead of trying to interview experts at length to capture their knowledge, today, someone in analytics would prefer to review the thousands of cases where the characteristics of the case, the decision by an expert and outcome was known. Then some kind of machine learning would search for the underlying patterns. In that way, the expert’s tacit understanding of the world would arise out of the analytics.
Nor does this cut out experts in the knowledge acquisition process. It just changes their role from being a memoirist. Instead, the experts can help kick off the building of the model and even assist in interpreting the results of the analytics.
Once the learning has begun, there is still much to learn (no pun intended) from the pioneers of this field. While they had great difficulty obtaining the knowledge – feeding the learning organization – they knew the rules of distributing that knowledge and making it useful. That is a lesson that today’s folks specializing in analytics could learn. Among these:
The importance of organizational culture
Leadership interest and support – especially for open discussion and experiment
Measurement (which would certainly provide grist for the analytics mill)
Systems thinking
For an illustration, see “Evidence in the learning organization” from the National Institutes of Health, in which these issues are focused on the medical profession.
If a marriage of the learning organization and knowledge management movements with the newer focus on analytics takes place, then all of those fields will improve and benefit.
What do a doctor suggesting a patient diet, an IT person asking a fellow worker to start using a new application, a mom asking a child to eat something or an analytics expert asking a business person to trust statistical insights all have in common? They are trying to get other people to do things.
But as the old saying goes, too often, this suggestion comes across as “do as I say, not as I do.” We’ve heard other variations on this theme, such as “doctor, heal thyself” or my favorite, “eat your own cooking”.
When I was a CIO, I used to tease some of the staff by pointing out they were all too willing to tell everyone else that they should use some new technology to do their jobs, but when it came to new technology in the IT world, the IT staff was the most resistant to change.
Among the most important changes that need to happen in many organizations today are those based on new insights about the business from analytics. But in big organizations, it is difficult to know how well those necessary changes are being adopted.
Analytics can help with this problem too. Analytics tools can help to figure out what message about a change is getting across and how well is the change being adopted? In which offices, regions, kinds of people?
Yet, it is rare for analytics folks to use their tools to help guide their own success in getting analytics to be adopted. Here, though, are two examples, the first about individuals and the second about organizations as a whole.
Individual Willingness To Change
A couple of years ago, the Netherlands branch of Deloitte created a Change Adoption Profiler (CAP) model of their clients’ employees based on the willingness to adopt changes. As they describe it:
“Imagine being able to predict who will adopt change, and how they will adopt it before the change has even occurred. At Deloitte, we have developed a data driven decision making method called the Change Adoption Profiler – it provides insights into your company’s attitude toward change and allows you to address it head on.
“The CAP uses a diagnostic survey based on personal characteristic and change attitudes. Unlike traditional questionnaires CAP combines this with behavioral data to understand the profiles that exist within the organization. The CAP provides reliable, fact-based analytics – provides client insights that support smart decision making, reveals risks and signals how to approach change at an early stage.”
Sadly, so far as I can tell from the public web, no other office of Deloitte is using this model in its work.
Organizational Analysis
Network analysis, especially social network analysis, is not a new focus of those in analytics. But, again, they don’t normally use network analysis to understand how well changes are being spread through an organization or business ecosystem.
One of the exceptions is the Danish change management consulting firm, Innovisor. They put particular emphasis on understanding the real organization – how and why people interact with each other – instead of relying solely or mostly on the official organization chart.
In his blog post, Henry Ward, CEO of eShares, writes at some length about his company’s use of this network analysis to determine who were the real influencers in the organization. They ended up identifying 9 employees, not necessarily executives, who influenced 70% of all employees directly and 100% through a second connection. A detailed presentation can be found at https://esharesinc.box.com/shared/static/8rrdq4diy3kkbsyxq730ry8sdhfnep60.pdf
Given the value of these kinds of examples, the problem of not eating your own lunch is especially interesting in the analytics field. Perhaps it is new enough that many of its advocates still have the zeal of being part of an early, growing religion and can’t see why others might resist their insights. But they would be convincing if they could show how, with analytics, they did their own jobs better – the job of getting their organizations to achieve the potential of their insights.
Last week, at the end of my class in Analytics and Leading Change, one of the required courses in Columbia University’s Masters Program in Applied Analytics, my students asked for books I’d recommend that provide more detail than we could cover in the course. It turns out that others are also interested in a good library of books about analytics from the viewpoint of an organization’s leaders.
You’ll see that these are not textbooks about analytics or machine learning techniques – there are plenty of those. Instead, this reading list is the next step for those folks who understand the techniques and now want the insights from their work to have an impact on and provide value to their world.
Although most of these books were published in the last decade, there are also some classics on the list going back fifty years. And I’ve chosen mostly popular books because frankly they are written in a compelling way that is accessible to all leaders.
With that introduction, here are my recommendations.
1. On the experience of doing analytics and seeing its impact:
Moneyball by Michael Lewis
The movie, Moneyball, starred Brad Pitt as the hero of the first and most storied use of analytics in professional baseball. For people in the field of analytics, what could be better than a movie about your skills helping the underdog. But like all movies, it tended to gloss over or exaggerate situations for the benefit of a good, simple plot.
The book that Lewis wrote originally is subtler and is a good case study of the human side of introducing analytics in a tradition-bound field. Tying it all up, his more recent book, The Undoing Project: A Friendship that Changed Our Minds, is the story of the collaboration between Kahneman (see below) and Tversky.
The Signal and The Noise: Why So Many Predictions Fail — But Some Don’t by Nate Silver
Nate Silver is probably the best-known analytics practitioner by those not in the business themselves, due to his work over the years, especially for the New York Times and in relation to high visibility elections. This is his review of the ups and downs in using analytics, offering lessons especially from sports and politics.
Victory Lab: The Secret Science of Winning Campaigns by Sasha Issenberg
Although sometimes a bit over the top and now five years old, it is a thorough description of the use of analytics in election campaigns. Election campaigns are good examples of analytics because they are both well-known and there is a huge amount of data concerning elections and the voters who determine their outcomes.
Dataclysm: Love, Sex, Race, and Identity — What Our Online Lives Tell Us about Our Offline Selves by Christian Rudder
The author is the co-founder and former analytics lead for OkCupid. Not surprisingly, much of the book is about dating choices, but he goes way beyond that to uncover insights about various social attitudes, including racism, from the large amount of data he had in his hands both at his former company and elsewhere.
How Not To Be Wrong: The Power Of Mathematical Thinking by Jordan Ellenberg
Since analytics is essentially a mathematical art, Ellenberg’s book about mathematical thinking is important preparation for the field. It also provides numerous examples of how to present quantitative insights in a way that non-experts would understand.
2. On expanding the normal range of analytics:
Unobtrusive Measures: Nonreactive Research in the Social Sciences by Eugene Webb, et al
I’ve added this fifty year old classic to the list because even in a world of big data we don’t necessarily have all the data we need, either in our computer systems or in the physical world. This book reminds us to observe indications of phenomenon that are not already available – such as the influence of an individual measured by the wear and tear on the entry to his/her office space. It also points out the need to always include metadata in our analysis since that is often revealing.
How to Measure Anything: Finding the Value of Intangibles in Business by Douglas Hubbard
Somewhat picking up the same theme, this book helps both the business executive and the analytics practitioner to be more creative in measurement, especially when it comes to things that people haven’t so far been able to offer good metrics for.
Connected: The Surprising Power of Social Networks and How They Shape Our Lives by Nicholas A. Christakis and James H. Fowler
This is a book about how social networks influence us in ways we hadn’t considered before. As they say: “How your friends’ friends’ friends affect everything you think, feel and do.” I suppose a good example is how their observation that you’ll gain weight by being connected to overweight people in a social network has itself become a meme. In its own way, this book is an interesting work of analytics.
Just as important is its elaboration of how to study social networks since an understanding of the network of influencers in any organization is essential to anyone who wants to change the behavior of the people in that organization.
Storytelling with Data: A Data Visualization Guide for Business Professionals by Cole Nussbaumer Knaflic
The author was part of Google’s analytics team, which is the analytics equivalent of working at the Vatican if you’re a Roman Catholic theologian. Her emphasis in on how to show the insights of analytics work and to tell a story about those insights. In a world of all kinds of data visualization tools and fads, her advice is clear and evidence-based.
3. On the way that the human mind perceives the insights of analytics and might or might change as a result:
Payoff: The Hidden Logic That Shapes Our Motivations by Dan Ariely
Professor Ariely, formerly of MIT and now at Duke, is one of the more creative experimenters in psychology and he quickly reviews both his own and others’ research results. The theme of this short book is that the payoff which often makes a difference in human behavior is not necessarily a financial reward and that sometimes financial incentives even backfire. This is important for leaders of change in organizations, particularly big corporations, to understand.
Thinking, Fast And Slow by Daniel Kahneman
I’ve written about the work of Nobel Prize winner and Princeton Professor Kahneman before, most recently in “What Do We Know About Change”. This describes what Kahneman has learned from a lifetime of research about thinking and decision making. His work on how people process – distort – quantitative statements is especially relevant to analytics experts who need understand the cognitive biases he describes.
Switch: How to Change Things When Change Is Hard by Chip Heath and Dan Heath
The Heath brothers, popular business writers, have done a good job in this book of explaining what’s been learned in recent psychological research – see Kahneman and Ariely, for instance – without dumbing it down so much that the key points are lost. In doing that well, they also provide the leader of change and analytics some good ideas on how to present their own results and getting their organizations to switch to a more analytics-oriented outlook.
4. On the strategic linkage between leading change and analytics
The Dance of Change by Peter Senge, et al
This is another classic that goes beyond the usual cookbook approach found in most books on “change management”. Yet, Senge and his colleagues anticipated the more recent approaches to change management which is about something more than just getting a single project done. For Senge, the goal he established was to help create learning organizations. While he does not focus on analytics, this book should particularly resonate with analytics professionals since they now have the tools to take that learning to new and more useful levels than in the past.
I could easily expand this list, as could many others, but this “baker’s dozen” books will provide a good rounded education to start.
Next week, I’m teaching the summer semester version of my Columbia University course called Analytics and Leading Change for the Master’s Degree program in Applied Analytics. While there are elective courses on change management in business and public administration schools, this combination of analytics and change is unusual. The course is also a requirement. Naturally, I’ve been why?
The general answer is that analytics and change are intertwined.
Successfully introducing analytics into an organization shares all the difficulties of introducing any new technology, but more so. The impact of analytics – if successful – requires change, often deep change that can challenge the way that executives have long thought about the effect of what they were doing.
A good, but early example of the impact of what we now call “big data”, goes back twenty-five years ago to the days before downloaded music.
Back then, the top 40 selections of music on the “air” were based on what radio DJs (or program directors) chose and, beyond that, the best information about market trends came from surveys of ad hoc observations by record store clerks. Those choices too emphasized new mainstream rock and pop music.
In 1991, in one of the earliest big data efforts in retail, a new company, SoundScan, came along and collected data from automated sales registers in music. What they found went against the view of the world that was then widely accepted –
and instead
old music, like Frank Sinatra, and genres others than rock were very popular.
A somewhat more recent example is the way that insights from analytics have challenged some of the traditional assumptions about motivation that are held by many executives and many staff in corporate human resource departments. Tom Davenport’s Harvard Business Review article in 2010 on “Competing on Talent Analytics” provides a good review of what can be learned, if executives are willing to learn from analytics.
The first, larger lesson is: If the leaders of analytics initiatives don’t understand the nature of the changes they are asking of their colleagues, then those efforts will end up being nice research reports and the wonderful insights generated by the analysts will disappear without impact or benefit to their organizations.
The other side of the coin and the second reason that analytics and change leadership are intertwined is a more positive one. Analytics leaders have a potential advantage over other “change agents” in understanding how to change an organization. They can use analytics tools to understand what they’re dealing with and thus increase the likelihood that the change will stick.
For instance, with the rise of social networks on the internet, network analytics methods have developed to understand how the individuals in a large group of people influence each other. Isn’t that also an issue in understanding the informal, perhaps the real, structure of an organization which the traditional organization charts don’t illuminate?
In another, if imperfect example, the Netherlands office of Deloitte created a Change Adoption Profiler to help leaders figure out the different reactions of people to proposed changes.
Unfortunately, leaders of analytics in many organizations too infrequently use their own tools to learn what they need to do and how well they are doing it. Pick your motto about this – “eat your own lunch (or dogfood)” or “doctor heal thyself” or whatever – but you get the point.
One of the more interesting technologies that has been developing is called affective computing. It’s about analyzing observations of human faces, voices, eye movements and the like to understand human emotions — what pleases or displeases people or merely catches their attention. It combines deep learning, analytics, sensors and artificial intelligence.
While interest in affective computing hasn’t been widespread, it may be nearing its moment in the limelight. One such indication is that the front page of the New York Times, a couple of days ago, featured a story about its use for television and advertising. The story was titled “For Marketers, TV Sets Are an Invaluable Pair of Eyes.”
But the companies that were featured in the Times article are not the only ones or the first ones to develop and apply affective computing. IBM published a booklet on the subject in 2001. Before that, in 1995, the term “affective computing” was coined by Professor Rosalind Picard of MIT, who also created the affective computing group in the MIT Media Lab.
In a video, “The Future of Story Telling”, she describes what is essentially the back story to the New York Times article. In no particular order, among other companies working with this technology today, there are Affectiva, Real Eyes, Emotient, Beyond Verbal, Sension, tACC, nVisio, CrowdEmotion, PointGraB, Eyeris, gestigon, Intel RealSense, SoftKinetic, Elliptic Labs, Microsoft’s VIBE Lab and Kairos.
Affectiva, which Professor Picard co-founded, offers an SDK that reads emotions of people at home or in the office just by using web cams. Here’s a video that shows their commercially available product at work: https://www.youtube.com/watch?v=mFrSFMnskI4
The
two previous products have obvious application to web marketing and
content. So much so, that some predict a future in which affective
technology creates an “emotion economy”.
But affective computing
has longer term applications, most especially in robotics. As human-like
robots, especially for an aging population in Asia, begin to be sold as
personal assistants and companions, they will need to have the kind of
emotional intelligence about humans that other human beings mostly have
already. That’s likely to be where we will see some of the most
impactful uses of affective computing.
Over the last couple of
years, Japan’s Softbank has developed Pepper, which they describe as a
“social robot” since it aims to recognize human emotion and shows its
own emotions. Here’s the French software company behind Pepper — https://www.youtube.com/watch?v=nQFgGS8AAN0
There
are others doing the same thing. At Nanyang Technological University,
Singapore, another social robot, called Nadine, is being developed. See https://www.youtube.com/watch?v=pXg33S3U_Oc
Both
these social robots and affective computing overall still needs much
development, but already you can sense the importance of this
technology.
As I’ve been going through articles and books for the course on Analytics and Leading Change that I’ll be teaching soon at Columbia University, I frequently read how leaders and other change agents need to overcome resistance to change. Whenever we aim to get things done and they don’t happen immediately, this is often the first explanation for the difficulty.
Resistance to change is a frequent complaint of anyone introducing a new technology or especially something as fundamental as the use of analytics in an organization.
The conflict that it implies can be compelling. You could make a best seller or popular movie out of that conflict, like that great story about baseball, analytics and change “Moneyball”.
This is an idea that goes very far back. Even Machiavelli, describing Renaissance politics, is often quoted on the subject:
“There is nothing more difficult to take in hand, more perilous to conduct, or more uncertain in its success, than to take the lead in the introduction of a new order of things. For the reformer has enemies in all those who profit by the old order, and only lukewarm defenders in all those who would profit by the new order, this lukewarmness arising partly from fear of their adversaries … and partly from the incredulity of mankind, who do not truly believe in anything new until they have had actual experience of it.”
It’s all awful if you’re the one trying to introduce the change and many have written about the problems they saw.
But is that word “resistance” misleading change agents? Going beyond the perspectives and anecdotes of change agents and business consultants, there has been over the last two decades some solid academic research on this subject. And, as often happens when we learn more, there have been some important subtleties lost in that phrase “resistance to change”.
In perhaps a refutation or an elaboration on Machiavelli’s famous quote, Dent and Goldberg report in “Challenging ‘Resistance to Change’” that:
“People do not resist change, per se. People may resist loss of status, loss of pay, or loss of comfort, but these are not the same as resisting change … Employees may resist the unknown, being dictated to, or management ideas that do not seem feasible from the employees’ standpoint. However, in our research, we have found few or no instances of employees resisting change … The belief that people do resist change causes all kinds of unproductive actions within organizations.”
Is what looks like resistance something more or something else?
More recently, University of Montreal Professor Céline Bareil wrote about the “Two Paradigms about Resistance to Change” in which she compared “the enemy of change” (traditional paradigm) to “a resource” (modern paradigm). She noted that:
“Instead of being interpreted as a threat, and the enemy of change, resistance to change can also be considered as a resource, and even a type of commitment on the part of change recipients.”
Making this shift in perspective is likely harder for change agents than the changes they expect of others. The three authors of “Resistance to Change: The Rest of the Story” describe the various ways that change agents themselves have biased perceptions. They say that blaming difficulties on resistance to change may be a self-serving and “potentially self-fulfilling label, given by change agents attempting to make sense of change recipients’ reactions to change initiatives, rather than a literal description of an objective reality.”
Indeed, they observe that the actions of change agents may not be merely unsuccessful, but counter-productive.
“Change agents may contribute to the occurrence of the very reactions they label as resistance through their own actions and inactions, such as communications breakdowns, the breach of agreements and failure to restore trust” as well as not listening to what is being said and learning from it.
There is, of course, a lot more to this story, which you can start to get into by looking at some of the links in this post. But hopefully this post has offered enough to encourage those of us who are leading change to take a step back, look at the situation differently and thus be able to succeed.
As
you will know from the news media, business executives and the techno-sphere,
we are in the age of big data and analytics.
(Disclosure: I too am part of this trend with my forthcoming course on
leading change in the Applied Analytics Master’s program at Columbia University.)
For
those of us who have been practitioners of analytics, this attention is long
overdue. But there is a certain naiveté
in the breathless stories we have all read and in many of the uses – really misuses
– of analytics that we see now.
Partly to provide a more mature understanding of analytics, Kaiser Fung,
the director of the Columbia program, has written an insightful book
titled “NumberSense”.
Filled with compelling examples, the book is a
general call for more sophistication in this age of big data. I like to
think of it as a warning that a superficial look at the numbers you
first see will not necessarily provide the most accurate picture, any
more than the first thing you see about an unpeeled onion tells you as
much as you can see once it is cut.
Continuing this theme in his recent book, “The End Of Average”, Todd
Rose has popularized the story of the Air Force’s misuse of averages and
rankings after World War II. He describes how the Air Force was faced
with an inexplicable series of accidents despite nothing being wrong
with the equipment or seemingly with the training of the pilots. The
Air Force had even gone to the effort of designing the cockpits to fit
the exact dimensions of the average pilot!
“In
the early 1950s, the U.S. air force measured more than 4,000 pilots on
140 dimensions of size, in order to tailor cockpit design to the
‘average’ pilot … [But] Out of 4,063 pilots, not a single airman fit
within the average range on all 10 dimensions. One pilot might have a
longer-than-average arm length, but a shorter-than-average leg length.
Another pilot might have a big chest but small hips. Even more
astonishing, Daniels discovered that if you picked out just three of the
ten dimensions of size — say, neck circumference, thigh circumference
and wrist circumference — less than 3.5 per cent of pilots would be
average sized on all three dimensions. Daniels’s findings were clear
and incontrovertible. There was no such thing as an average pilot.
If you’ve designed a cockpit to fit the average pilot, you’ve actually
designed it to fit no one.”
Rose criticizes the very popular
one-dimension rankings and calls for an understanding of the full
complexity, the multi-dimensional nature of human behavior and
performance. As a Harvard Professor of Education, he puts special
emphasis on the misleading rankings that every student faces.
He shows three ways that these averages can mislead by not recognizing that:
The
one number used to rank someone actually represents multiple dimensions
of skills, personality and the like. Two people can have the same
score, but actually have a very different set of attributes.
Behavior and skill change depending upon context.
The
path to even the same endpoint can be different for two people. While
they may look the same when they get there, watching their progress
shows a different picture. He provides, as an example, the various
patterns of infants learning to walk. Eventually, they all do learn,
but many babies do not follow any standard pattern of doing so.
It
is not too difficult to take this argument back to Michael Lewis’s
portrayal in Moneyball of the way that the Oakland A’s put together a
successful roster by not selecting those who looked like star baseball
athletes – a uni-dimensional, if very subjective, ranking.
Let’s
hope that as big data analytics mature, there are more instances of
Moneyball sophistication and less of the academic rankings that Rose
criticizes.
The accumulation of data is increasing fast – from wearables, the widespread deployment of sensors in physical locations and the ever increasing use of the Internet by people.
And someone somehow has to figure it all out. As a result, data scientists are in demand and analytics is a hot new field of study. On top of long standing statistical methods, there has been impressive progress recently in machine learning, artificial intelligence and new computer system architectures.
Yet, the use of analytics itself has not had as great an impact on many organizations as the data scientists have hoped. Some of the failures of analytics were really failures of implementation.
Perhaps the most public of these was the great Netflix million-dollar prize for a new recommendation engine. From a purely technical viewpoint, the winning team did exactly what was asked for – create a significantly better engine than the one Netflix had been using. Nevertheless, Netflix ended up not using their work. That’s an issue of implementation and integrating the product of analytics into an organization.
Being able to predict behavior or even generate new insights from all this data is one thing. As with Netflix, having people and organizations actually use that knowledge is another. Like many other new technologies, adoption is as much a question of managing change as it is developing the technology itself.
This surely bothers some data scientists. After all, they have a better mousetrap – why aren’t people using it? Being able to think quantitatively, they can prepare quite convincing business cases with impressive ROI statistics and yet even that isn’t enough to get executives to budge. But changing an organization isn’t simple no matter how good your arguments.
Despite this background, there has been very little overlap between the courses that prepare data scientists and the courses that prepare change agents in organizations.
Later this year, I’ll be doing something to help align these two fields to improve the success of both. I’ll be teaching an online course on analytics and leading change. It will be part of Columbia University’s new executive graduate program in Applied Analytics.
We’ll be reviewing what is known about successfully introducing changes into an organization from the classics on the subject that were written as much as twenty years ago to more recent research. The course will, of course, help its students understand how to get an analytics initiative started. More important, it will focus on how to sustain, over the long run, both analytics and the changes it informs.
Thinking about the long run, there are three facets of the relationship between analytics and organizational change.
The use of analytics as a part of everyday decision making and the rules of operation in a business – this is the obvious first thing everyone thinks of.
The use of analytics to help better implement the changes that its insights imply – a kind of a meta-analysis.
The continuing interaction between analytics and change to finally achieve the long desired goal of an organization that learns how to continually optimize itself – this is something of great strategic value to any business.
As the course develops, I’ll be posting more about each topic.
Last week, on April 16th, the Knowledge Society Forum of the Eurocities group held its Beyond Data event in Eindhoven, the Netherlands. The members of the KSF consists out of more than 50 policy makers focused on Open Data, from Europe. They were joined by many other open data experts and advocates.
I led off with the keynote presentation. The theme was simple: we need to go beyond merely opening (i.e., releasing) public data and there are a variety of new technologies that will make the Open Data movement more useful to the general public.
Since I was speaking in my role as Senior Fellow of the Intelligent Community Forum (ICF), I drew a parallel between that work and the current status of Open Data. I pointed out that ICF has emphasized that an “intelligent city” is much more than a “smart city” with technology controlling its infrastructure. What makes a community intelligent is if and how it uses that technology foundation to improve the experience of living there.
Similarly, to make the open data movement relevant to citizens, we need to go beyond merely releasing public data. Even Hackathons and the encouragement of app developers has its limits in part because developers in private companies will try to find some way to monetize their work, but not all useful public problems have profit potential.
To create this value means focusing on data of importance to people (not just what’s easy to deliver), undertaking data analytics, following up with actions that have real impact on policies and programs and especially, engaging citizen in every step of the open data initiative.
I pointed out how future technology trends will improve every city’s use of its data in three ways:
1. Data collection, integration and quality
2. Visualization, anywhere it is needed
3. Analytics of the data to improve public policies and programs
For example, the inclusion of social data (like sentiment analysis) and the Internet of Things can be combined with data already collected by the government to paint a much richer picture of what is going on in a city. In addition to drones, iBeacon, visual analyzers (like Placemeter), there are now also inexpensive, often open source, sensor devices that the public can purchase and use for more data collection.
Of course, all this data needs a different kind of management than businesses have used in the past. So I pointed out NoSQL database management systems and Dat for real time data flow. Some of the most interesting analytics is based on the merger of data from multiple sources, which poses additional difficulties that are beginning to be overcome through linked data and the new geospatial extension of the semantic web, GeoSPARQL.
If this data – and the results of its analysis – are to be useful, especially in real time, then data visualization needs to be everywhere. That includes using augmented reality and even projecting results on surfaces, much like TransitScreen does.
And if all this data is to be useful, it must be analyzed so I discussed the key role of predictive analytics in going beyond merely releasing data. But I emphasized the way that residents of a city can help in this task and cited the many people already involved in Zooniverse. There are even tools to help people overcome their statistical immaturity, as you can see on Public Health Ontario.
Finally, the data can also be used by people to help envision – or re-envision – their cities through tools like Betaville.
Public officials have to go beyond merely congratulating themselves on being transparent by releasing data. They need to take advantage of these technological developments and shift their focus to making the data useful to their residents – all in the overriding goal of improving the quality of life for their residents.
By now, lots of
people have heard about Big Data, but the message often comes across as another
corporate marketing phrase and a message with multiple meanings. That may be because people also hear from
corporate executives who eagerly anticipate big new revenues from the Big Data
world.
However, I
suspect that most people don’t know what Big Data experts are talking about,
what they’re doing, what they believe about the world, and the issues arising
from their work.
Although it was
originally published in 2013, the book “Big Data: A Revolution That Will
Transform How We Live, Work, And Think” by Viktor Mayer-Schönberger and Kenneth Cukier is perhaps the best recent
in-depth description of the world of Big Data.
For people like
me, with an insatiable curiosity and good analytical skills, having access to
lots of data is a treat. So I’m very
sympathetic to the movement. But like
all such movements, the early advocates can get carried away with their
enthusiasm. After all, it makes you feel
so powerful as I recall some bad sci fi movies.
Here then is a
summary of some key elements of Big Data thinking – and some limits to that
thinking.
Causation and
Correlation
When
presented with the result of some analysis, we’ve often been reminded that
“correlation is not causation”, implying we know less than we think if all we
have is a correlation.
For
many Big Data gurus, correlation is better than causation – or at least finding
correlations is quicker and easier than testing a causal model, so it’s not
worth putting the effort into building that model of the world. They say that causal models may be an outmoded
idea or as Mayer-Schönberger
and Cukier say, “God is dead”. They add
that “Knowing what, rather than why, is good enough” – good enough, at least,
to try to predict things.
This
isn’t the place for a graduate school seminar on the philosophy of science, but
there are strong arguments that models are still needed whether we live in a
world of big data or not.
All The Data, Not Just
Samples
Much
of traditional statistics dealt with the issue of how to draw conclusions about
the whole world when you could only afford to take a sample. Big data experts say that traditional
statistics’ focus is a reflection of an outmoded era of limited data.
Indeed, an example is a 1975 textbook that
was titled “Data Reduction: Analysing and Interpreting Statistical Data”. While
Big Data provides lots more opportunity for analysis, it doesn’t overcome all
the weaknesses that have been associated with statistical analysis and
sampling. There can still be measurement
error. Big Data advocates say the sheer
volume of data reduces the necessity of being careful about measurement error,
but can’t there still systematic error?
Big Data gurus say that they include all the data, not just a sample. But, in a way, that’s clearly an
overstatement. For example, you can
gather all the internal records a company has about the behavior and breakdowns
of even millions of devices it is trying to keep track of. But, in fact, you may not have collected all
the relevant data. It may also be a
mistake to assume that what is observed about even all people today will
necessarily be the case in the future – since even the biggest data set today
isn’t using tomorrow’s data.
More Perfect Predictions
The
Big Data proposition is that massive volumes of data allows for almost perfect
predictions, fine grain analysis and can almost automatically provide new
insights. While these fine grain
predictions may indicate connections between variables/factors that we hadn’t
thought of, some of those connections may be spurious. This is an extension of the issue of
correlation versus causation because there is likely an increase in spurious
correlations as the size of the data set increases.
If
Netflix recommends movies you don’t like, this isn’t a big problem. You just ignore them. In the public sector, when this approach to
predicting behavior leads to something like racial profiling, it raises legal
issues.
It
has actually been hard to find models that achieve even close to perfect
predictions – even the well-known stories about how Farecast predicted the best
time to buy air travel tickets or how Google searches predicted flu outbreaks. For a more general review of these
imperfections, read Kaiser Fung’s “Why
Websites Still Can’t Predict Exactly What You Want” published in Harvard
Business Review last year.
Giving It All Away
Much
of the Big Data movement depends upon the use of data from millions – billions?
– of people who are making it available unknowingly, unintentionally or at
least without much consideration.
Slowly,
but surely, though, there is a developing public policy issue around who has
rights to that data and who owns it.
This past November’s Harvard
Business Review – hardly a radical fringe journal – had an article that
noted the problems if companies continue to assume that they own the
information about consumers’ lives. In
that article, MIT Professor Alex Pentland proposes a “New Deal On Data”.
So Where Does This Leave
Us?
Are
we much better off and learning much more with the availability of Big Data,
instead of samples of data, and the related ability of inexpensive computers
and software to handle this data?
Absolutely, yes!
As
some of the big egos of Big Data claim, is Big Data perfect enough to withhold
some skepticism about its results? Has Big Data become the omniscient god? – Not
quite yet.
[Published 6/18/2011 and originally posted for government leaders, July 6, 2009]
My last posting was about the “goldmine” that exists in the information your government collects every day. It’s a goldmine because this data can be analyzed to determine how to save money by learning what policies and programs work best. Some governments have the internal skills to do this kind of sophisticated analysis or they can contract for those skills. But no government – not even the US Federal government – has the resources to analyze all the data they have.
What can you do about that? Maybe there’s an answer in a story about real gold mining from the authors of the book “Wikinomics”[1]:
A few years back, Toronto-based gold mining company Goldcorp was in trouble. Besieged by strikes, lingering debts, and an exceedingly high cost of production, the company had terminated mining operations…. [M]ost analysts assumed that the company’s fifty-year old mine in Red Lake, Ontario, was dying. Without evidence of substantial new gold deposits, Goldcorp was likely to fold. Chief Executive Officer Rob McEwen needed a miracle.
Frustrated that his in-house geologists couldn’t reliably estimate the value and location of the gold on his property … [he] published his geological data on the Web for all to see and challenged the world to do the prospecting. The “Goldcorp Challenge” made a total of $575,000 in prize money available to participants who submitted the best methods and estimates. Every scrap of information (some 400 megabytes worth) about the 55,000 acre property was revealed on Goldcorp’s Web site.
News of the contest spread quickly around the Internet and more than 1,000 virtual prospectors from 50 countries got busy crunching the data. Within weeks, submissions from around the world were flooding into Goldcorp headquarters. There were entries from graduate students, management consultants, mathematicians, military officers, and a virtual army of geologists. “We had applied math, advanced physics, intelligent systems, computer graphics, and organic solutions to inorganic problems. There were capabilities I had never seen before in the industry,” says McEwen. “When I saw the computer graphics, I almost fell out of my chair.”
The contestants identified 110 targets on the Red Lake property, more than 80% of which yielded substantial quantities of gold. In fact, since the challenge was initiated, an astounding 8 million ounces of gold have been found – worth well over $3 billion. Not a bad return on a half million dollar investment.
You probably won’t be able to offer a prize to analysts, although you might offer to share some of the savings that result from doing things better. But, since the public has an interest in seeing its government work better, unlike a private corporation, maybe you don’t have to offer a prize.And there are many examples on the Internet where people are willing to help out without any obvious monetary reward.
Certainly not everyone, but enough people might be interested in the data to take a shot of making sense of it – students or even college professors looking for research projects, retired statisticians, the kinds of folks who live to analyze baseball statistics, and anyone who might find this a challenge.
The Obama administration and its new IT leaders have made a big deal about putting its data on the Web. There are dozens of data sets on the Federal site data.gov[2], obviously taking care to deal with issues of individual privacy and national security. Although their primary interest is in transparency of government, now that the data is there, we’ll start to see what people out there learn from all that information. Alabama[3] and the District of Columbia, among others, have started to do the same thing.
You can benefit a lot more, if you too make your government’s data available on the web for analysis. Then your data, perhaps combined with the Federal data and other sources on the web, can provide you with an even better picture of how to improve your government – better than just using your own data alone.
[Note: This was originally posted on a blog for government leaders, July 6, 2009]
My last posting was about the “goldmine” that exists in the information your government collects every day. It’s a goldmine because this data can be analyzed to determine how to save money by learning what policies and programs work best. Some governments have the internal skills to do this kind of sophisticated analysis or they can contract for those skills. But no government – not even the US Federal government – has the resources to analyze all the data they have.
What can you do about that? Maybe there’s an answer in a story about real gold mining from the authors of the book “Wikinomics”[1]:
A few years back, Toronto-based gold mining company Goldcorp was in trouble. Besieged by strikes, lingering debts, and an exceedingly high cost of production, the company had terminated mining operations…. [M]ost analysts assumed that the company’s fifty-year old mine in Red Lake, Ontario, was dying. Without evidence of substantial new gold deposits, Goldcorp was likely to fold. Chief Executive Officer Rob McEwen needed a miracle.
Frustrated that his in-house geologists couldn’t reliably estimate the value and location of the gold on his property … [he] published his geological data on the Web for all to see and challenged the world to do the prospecting. The “Goldcorp Challenge” made a total of $575,000 in prize money available to participants who submitted the best methods and estimates. Every scrap of information (some 400 megabytes worth) about the 55,000 acre property was revealed on Goldcorp’s Web site.
News of the contest spread quickly around the Internet and more than 1,000 virtual prospectors from 50 countries got busy crunching the data. Within weeks, submissions from around the world were flooding into Goldcorp headquarters. There were entries from graduate students, management consultants, mathematicians, military officers, and a virtual army of geologists. “We had applied math, advanced physics, intelligent systems, computer graphics, and organic solutions to inorganic problems. There were capabilities I had never seen before in the industry,” says McEwen. “When I saw the computer graphics, I almost fell out of my chair.”
The contestants identified 110 targets on the Red Lake property, more than 80% of which yielded substantial quantities of gold. In fact, since the challenge was initiated, an astounding 8 million ounces of gold have been found – worth well over $3 billion. Not a bad return on a half million dollar investment.
You probably won’t be able to offer a prize to analysts, although you might offer to share some of the savings that result from doing things better. But, since the public has an interest in seeing its government work better, unlike a private corporation, maybe you don’t have to offer a prize.And there are many examples on the Internet where people are willing to help out without any obvious monetary reward.
Certainly not everyone, but enough people might be interested in the data to take a shot of making sense of it – students or even college professors looking for research projects, retired statisticians, the kinds of folks who live to analyze baseball statistics, and anyone who might find this a challenge.
The Obama administration and its new IT leaders have made a big deal about putting its data on the Web. There are dozens of data sets on the Federal site data.gov[2], obviously taking care to deal with issues of individual privacy and national security. Although their primary interest is in transparency of government, now that the data is there, we’ll start to see what people out there learn from all that information. Alabama[3] and the District of Columbia, among others, have started to do the same thing.
You can benefit a lot more, if you too make your government’s data available on the web for analysis. Then your data, perhaps combined with the Federal data and other sources on the web, can provide you with an even better picture of how to improve your government – better than just using your own data alone.
[This was originally posted on the web on June 15, 2009 for elected executives of governments.]
Every day, the employees of your government follow the same routine.
They have a stack of problems, applications, forms and the like in their inbox. It may be a real, old-fashioned inbox with lots of paper or the computer-based equivalent. Doing the best they can, they then work through the pile and, we hope, with wisdom and efficiency, they process the incoming tasks and then move them to the outbox. As far as many employees are concerned, their work is done when the thing is put in the outbox.
However, for the people who run the government, this represents more than a ledger of what came in and what went out. It is a gold mine of information. Especially because of all the automation that has been put in place in government agencies, it is also an easily accessible gold mine.
Unfortunately, this gold mine is often ignored. But if that data is analyzed, you will discover the patterns that can help you improve government programs and policies. Consider two examples, from very different areas, of what statistical analysis of that data can tell you:
What kinds of programs have worked best for which kinds of prisoners? (This knowledge can be used to come up with better treatment and assignment of prisoners at intake.)
Who has used the public golf courses at what times of the week and day? (This can identify where you might want to offer new programs targeted at particular groups of residents to even out usage during the day and get more golf fees.)
In 2007, Professor Ian Ayres wrote a book, “SuperCrunchers: Why Thinking-By-Numbers Is The New Way To Be Smart”, in which he described how various organizations are using statistical analysis to dramatically improve their performance.
One of its chapters, “Government By Chance”, provides public sector examples and offers an interesting idea.
Imagine a world where people looked to the IRS as a source for useful information. The IRS could tell a small business that it might be spending too much on advertising or tell an individual that the aver age taxpayer in her income bracket gave mote to charity or made a larger IRA contribution. Heck, the IRS could probably produce fairly accurate estimates about the probability that small businesses (or even marriages) would fail. In fact, I’m told that Visa already does predict the probability of divorce based on credit card purchases (so that it can make better predictions of default risk). Of course, this is all a bit Orwellian. I might not particularly want to get a note from the IRS saying my marriage is at risk. But I might at least want the option of having the government make predictions about various aspects of my life. Instead of thinking of the IRS as solely a taker, we might also think of it as an information provider. We could even change its name to the “Information & Revenue Service".
This is yet another example, though, of moving the public sector from a transactional view of citizens to something more helpful.While even the author admits the IRS example is a scary, there are other possibilities that are not scary and that your residents would like.
The use of the data the government collects for better policy and better service to citizens is what I call “learning how to drive the government” because it is different from the usual fad and fashion approach to policy.
Too often policy debates are like a driver in a car who cannot see outside the windows. So the driver keeps going until the car hits a wall, at which point the usual reaction is to go in the opposite direction until the same thing happens again. This accounts for the feeling of a pendulum swinging in public policy debates, rather than real learning occurring.
When everyday data is analyzed, it is like being able to look out the windows and figure out what direction to drive.
[Re-published 5/18/2011. This was originally posted on the web on June 15, 2009 for elected executives of governments.]
Every day, the employees of your government follow the same routine.
They have a stack of problems, applications, forms and the like in their inbox. It may be a real, old-fashioned inbox with lots of paper or the computer-based equivalent. Doing the best they can, they then work through the pile and, we hope, with wisdom and efficiency, they process the incoming tasks and then move them to the outbox. As far as many employees are concerned, their work is done when the thing is put in the outbox.
However, for the people who run the government, this represents more than a ledger of what came in and what went out. It is a gold mine of information. Especially because of all the automation that has been put in place in government agencies, it is also an easily accessible gold mine.
Unfortunately, this gold mine is often ignored. But if that data is analyzed, you will discover the patterns that can help you improve government programs and policies. Consider two examples, from very different areas, of what statistical analysis of that data can tell you:
What kinds of programs have worked best for which kinds of prisoners? (This knowledge can be used to come up with better treatment and assignment of prisoners at intake.)
Who has used the public golf courses at what times of the week and day? (This can identify where you might want to offer new programs targeted at particular groups of residents to even out usage during the day and get more golf fees.)
In 2007, Professor Ian Ayres wrote a book, “SuperCrunchers: Why Thinking-By-Numbers Is The New Way To Be Smart”, in which he described how various organizations are using statistical analysis to dramatically improve their performance.
One of its chapters, “Government By Chance”, provides public sector examples and offers an interesting idea.
“Imagine a world where people looked to the IRS as a source for useful information. The IRS could tell a small business that it might be spending too much on advertising or tell an individual that the aver age taxpayer in her income bracket gave mote to charity or made a larger IRA contribution. Heck, the IRS could probably produce fairly accurate estimates about the probability that small businesses (or even marriages) would fail. In fact, I’m told that Visa already does predict the probability of divorce based on credit card purchases (so that it can make better predictions of default risk). Of course, this is all a bit Orwellian. I might not particularly want to get a note from the IRS saying my marriage is at risk. But I might at least want the option of having the government make predictions about various aspects of my life. Instead of thinking of the IRS as solely a taker, we might also think of it as an information provider. We could even change its name to the Information & Revenue Service”.
This is yet another example, though, of moving the public sector from a transactional view of citizens to something more helpful. While even the author admits the IRS example is a scary, there are other possibilities that are not scary and that your residents would like.
The use of the data the government collects for better policy and better service to citizens is what I call “learning how to drive the government” because it is different from the usual fad and fashion approach to policy.
Too often policy debates are like a driver in a car who cannot see outside the windows. So the driver keeps going until the car hits a wall, at which point the usual reaction is to go in the opposite direction until the same thing happens again. This accounts for the feeling of a pendulum swinging in public policy debates, rather than real learning occurring.
When everyday data is analyzed, it is like being able to look out the windows and figure out what direction to drive.
