Curiosity Is Not Intrinsically Good

Posted

Interesting reflections in Scientific American on morbid curiosity – that we are driven by our curiosity, sometimes even when we actually know that there is a strong likelihood it will hurt us. In the article, as the title implies, this is portrayed as a bad thing. I disagree.

“The drive to discover is deeply ingrained in humans, on par with the basic drives for food or sex, says Christopher Hsee of the University of Chicago, a co-author of the paper. Curiosity is often considered a good instinct—it can lead to new scientific advances, for instance—but sometimes such inquiry can backfire. “The insight that curiosity can drive you to do self-destructive things is a profound one,” says George Loewenstein, a professor of economics and psychology at Carnegie Mellon University who has pioneered the scientific study of curiosity.”

Bub in a boxThis is not exactly a novel, nor a profound insight: we even have a popular proverb for it that I mention to my cats on an almost daily basis. They don’t listen. 

There is a strong relationship between curiosity and the desire for competence: a need to know how things work, how to do something we cannot yet do, why things are the way they are, where our limits lie, how to become more capable of acting in the world. From an evolutionary perspective we are curious with a purpose. It allows us to make effective use of our environment, to become competent within it. This is really good for survival so, of course, it is selected for. That it sometimes drives us to do things that harm us is actually a very positive feature, as long as it is balanced with a sufficient level of caution and the harm it causes is not too great. It helps us to know what to avoid, as well as what is useful to us. It also helps us to be more adaptable to bad things that we cannot avoid. It makes us more flexible, and lets us both know and extend our limits.

The first experiment described here involved people playing with pens even knowing that some were novelty items that would give them an electric shock. I’m not sure why the researchers mixed in some harmless pens in this because, even when pain is an absolute certainty, curiosity can drive us to experience it. I have long used electrostatic zappers that are designed to alleviate the itch in mosquito bites by administering a sharp and slightly painful shock to the skin. I have yet to meet a single child and have met very few adults that did not want to try it out on their own skin, regardless of whether they had any bites, in the full and certain knowledge that it would hurt. This is described in the article as self-destructive curiosity but I don’t think that’s right at all. If subjects had been convincingly warned that some pens would kill or maim them, then I am quite certain that very few would have played with them (some might, of course – evolution thrives on variation and, in some environments, high-risk strategies might pay off). But being curious about what kind of pain it might cause is really just a way of discovering or achieving competence, of discovering how we cope with this kind of shock, of testing hypotheses about ourselves and the environment, as well as finding out whether such joke pens actually work as advertised. This is potentially useful information: it will make you less likely to be a victim of a practical joke, or perhaps inspire you to perform one more effectively. Either way, it’s probably not a big thing in the grand scheme of things but, then again, very few learning experiences are. The value is more about how we integrate and connect such experiences.

The article describes another experiment in which participants were encouraged to predict their feelings after being shown an unpleasant image. Those so primed were less likely to choose to see it. Again, this makes sense in the light of what we already know. We are curious with a purpose – to learn – so, if we reflect a bit on what we have already learned, then it might dull our curiosity to experience something bad again. That’s potentially useful. I’m not sure that it is always a good thing, though. I happen to like, say, some horror movies that disgust me, or comedies that rely on discomfort for their humour. In fact, the anticipation of fear or disgust is often one of the main things that drives their plots and keeps my eyes glued to them. If the zombie apocalypse comes, I will be totally prepared. It also prepares me better for things that are going to really upset me. Likewise for funfair rides, sailing on a breezy day, exercising until it hurts, eating hot chili, or struggling with difficult deadlines.

So while, yes, we absolutely should learn from experience, we also need to remember that it can lead us into fixed ways of thinking that can, when conditions change, be less adaptable and adaptive. There is an ever-shifting balance between fear and curiosity that we need to embrace, perhaps especially when curiosity leads to the likelihood of something unpleasant (though not too unpleasant) happening. And, even when the danger is great, there are also risks that are sometimes worth taking. ‘What if..?’ is one of the most powerful phrases in any language.

Address of the bookmark: http://www.scientificamerican.com/article/curiosity-is-not-intrinsically-good/

Cocktails and educational research

Posted , updated on

A lot of progress has been made in medicine in recent years through the application of cocktails of drugs. Those used to combat AIDS are perhaps the most well-known, but there are many other applications of the technique to everything from lung cancer to Hodgkin’s lymphoma. The logic is simple. Different drugs attack different vulnerabilities in the pathogens etc they seek to kill. Though evolution means that some bacteria, viruses or cancers are likely to be adapted to escape one attack, the more different attacks you make, the less likely it will be that any will survive.

Simulated learningUnfortunately, combinatorial complexity means this is not a simply a question of throwing a bunch of the best drugs of each type together and gaining their benefits additively. I have recently been reading John H. Miller’s ‘A crude look at the whole: the science of complex systems in business, life and society‘ which is, so far, excellent, and that addresses this and many other problems in complexity science. Miller uses the nice analogy of fashion to help explain the problem: if you simply choose the most fashionable belt, the trendiest shoes, the latest greatest shirt, the snappiest hat, etc, the chances of walking out with the most fashionable outfit by combining them together are virtually zero. In fact, there’s a very strong chance that you will wind up looking pretty awful. It is not easily susceptible to reductive science because the variables all affect one another deeply. If your shirt doesn’t go with your shoes, it doesn’t matter how good either are separately. The same is true of drugs. You can’t simply pick those that are best on their own without understanding how they all work together. Not only may they not additively combine, they may often have highly negative effects, or may prevent one another being effective, or may behave differently in a different sequence, or in different relative concentrations. To make matters worse, side effects multiply as well as therapeutic benefits so, at the very least, you want to aim for the smallest number of compounds in the cocktail that you can get away with. Even were the effects of combining drugs positive, it would be premature to believe that it is the best possible solution unless you have actually tried them all. And therein lies the rub, because there are really a great many ways to combine them.

Miller and colleagues have been using the ideas behind simulated annealing to create faster, better ways to discover working cocktails of drugs. They started with 19 drugs which, a small bit of math shows, could be combined in 2 to the power of 19 different ways – about half a million possible combinations (not counting sequencing or relative strength issues). As only 20 such combinations could be tested each week, the chances of finding an effective, let alone the best combination, were slim within any reasonable timeframe. Simplifying a bit, rather than attempting to cover the entire range of possibilities, their approach finds a local optimum within one locale by picking a point and iterating variations from there until the best combination is found for that patch of the fitness landscape. It then checks another locale and repeats the process, and iterates until they have covered a large enough portion of the fitness landscape to be confident of having found at least a good solution: they have at least several peaks to compare. This also lets them follow up on hunches and to use educated guesses to speed up the search. It seems pretty effective, at least when compared with alternatives that attempt a theory-driven intentional design (too many non-independent variables), and is certainly vastly superior to methodically trying every alternative, inasmuch as it is actually possible to do this within acceptable timescales.

The central trick is to deliberately go downhill on the fitness landscape, rather than following an uphill route of continuous improvement all the time, which may simply get you to the top of an anthill rather than the peak of Everest in the fitness landscape. Miller very effectively shows that this is the fundamental error committed by followers of the Six-Sigma approach to management, an iterative method of process improvement originally invented to reduce errors in the manufacturing process: it may work well in a manufacturing context with a small number of variables to play with in a fixed and well-known landscape, but it is much worse than useless when applied in a creative industry like, say, education, because the chances that we are climbing a mountain and not an anthill are slim to negligible. In fact, the same is true even in manufacturing: if you are just making something inherently weak as good as it can be, it is still weak. There are lessons here for those that work hard to make our educational systems work better. For instance, attempts to make examination processes more reliable are doomed to fail because it’s exams that are the problem, not the processes used to run them. As I finish this while listening to a talk on learning analytics, I see dozens of such examples: most of the analytics tools described are designed to make the various parts of the educational machine work ‘ better’, ie. (for the most part) to help ensure that students’ behaviour complies with teachers’ intent. Of course, the only reason such compliance was ever needed was for efficient use of teaching resources, not because it is good for learning. Anthills.

This way of thinking seems to me to have potentially interesting applications in educational research. We who work in the area are faced with an irreducibly large number of recombinable and mutually affective variables that make any ethical attempt to do experimental research on effectiveness (however we choose to measure that – so many anthills here) impossible. It doesn’t stop a lot of people doing it, and telling us about p-values that prove their point in more or less scupulous studies, but they are – not to put too fine a point on it – almost always completely pointless.  At best, they might be telling us something useful about a single, non-replicable anthill, from which we might draw a lesson or two for our own context. But even a single omitted word in a lecture, a small change in inflection, let alone an impossibly vast range of design, contextual, historical and human factors, can have a substantial effect on learning outcomes and effectiveness for any given individual at any given time. We are always dealing with a lot more than 2 to the power of 19 possible mutually interacting combinations in real educational contexts. For even the simplest of research designs in a realistic educational context, the number of possible combinations of relevant variables is more likely closer to 2 to the power of 100 (in base 10 that’s  1,267,650,600,228,229,401,496,703,205,376). To make matters worse, the effects we are looking for may sometimes not be apparent for decades (having recombined and interacted with countless others along the way) and, for anything beyond trivial reductive experiments that would tell us nothing really useful, could seldom be done at a rate of more than a handful per semester, let alone 20 per week. This is a very good reason to do a lot more qualitative research, seeking meanings, connections, values and stories rather than trying to prove our approaches using experimental results. Education is more comparable to psychology than medicine and suffers the same central problem, that the general does not transfer to the specific, as well as a whole bunch of related problems that Smedslund recently coherently summarized. The article is paywalled, but Smedlund’s abstract states his main points succinctly:

“The current empirical paradigm for psychological research is criticized because it ignores the irreversibility of psychological processes, the infinite number of influential factors, the pseudo-empirical nature of many hypotheses, and the methodological implications of social interactivity. An additional point is that the differences and correlations usually found are much too small to be useful in psychological practice and in daily life. Together, these criticisms imply that an objective, accumulative, empirical and theoretical science of psychology is an impossible project.”

You could simply substitute ‘education’ for ‘psychology’ in this, and it would read the same. But it gets worse, because education is as much about technology and design as it is about states of mind and behaviour, so it is orders of magnitude more complex than psychology. The potential for invention of new ways of teaching and new states of learning is essentially infinite. Reductive science thus has a very limited role in educational research, at least as it has hitherto been done.

But what if we took the lessons of simulated annealing to heart? I recently bookmarked an approach to more reliable research suggested by the Christensen Institute that might provide a relevant methodology. The idea behind this is (again, simplifying a bit) to do the experimental stuff, then to sweep the normal results to one side and concentrate on the outliers, performing iterations of conjectures and experiments on an ever more diverse and precise range of samples until a richer, fuller picture results. Although it would be painstaking and longwinded, it is a good idea. But one cycle of this is a bit like a single iteration of Miller’s simulated annealing approach, a means to reach the top of one peak in the fitness landscape, that may still be a low-lying peak. However if, having done that, we jumbled up the variables again and repeated it starting in a different place, we might stand a chance of climbing some higher anthills and, perhaps, over time we might even hit a mountain and begin to have something that looks like a true science of education, in which we might make some reasonable predictions that do not rely on vague generalizations. It would either take a terribly long time (which itself might preclude it because, by the time we had finished researching, the discipline will have moved somewhere else) or would hit some notable ethical boundaries (you can’t deliberately mis-teach someone), but it seems more plausible than most existing techniques, if a reductive science of education is what we seek.

To be frank, I am not convinced it is worth the trouble. It seems to me that education is far closer as a discipline to art and design than it is to psychology, let alone to physics. Sure, there is a lot of important and useful stuff to be learned about how we learn: no doubt about that at all, and a simulated annealing approach might speed up that kind of research. Painters need to know what paints do too. But from there to prescribing how we should therefore teach spans a big chasm that reductive science cannot, in principle or practice, cross. This doesn’t mean that we cannot know anything: it just means it’s a different kind of knowledge than reductive science can provide. We are dealing with emergent phenomena in complex systems that are ontologically and epistemologically different from the parts of which they consist. So, yes, knowledge of the parts is valuable, but we can no more predict how best to teach or learn from those parts than we can predict the shape and function of the heart from knowledge of cellular organelles in its constituent cells. But knowledge of the cocktails that result – that might be useful.

 

 

Oh yes, that's why I left

Posted

St George Cross (Wikipedia)England is a weird, sad, angry little country, where there is now unequivocal evidence that over half the population – mainly the older ones – believe that experts know nothing, and that foreigners (as well as milllions of people born there with darker than average skins) are evil. England is a place filled with drunkenness and random violence, where it’s not safe to pass a crowd of teenagers – let alone a crowd of football supporters – on a street corner, where you cannot hang Xmas decorations outside for fear of losing them, where your class still defines you forever, where whinging is a way of life, where kindness is viewed with suspicion, where barbed wire fences protect schools from outsiders (or vice versa – hard to fathom), where fuckin‘ is a punctuation mark to underline what follows, not an independent word. It’s a nation filled with fierce and inhospitable people, as Horace once said, and it always has been. For all the people and places that I love and miss there, for all its very many good people and slowly vanishing places that are not at all like that, for all its dark and delicious humour, its eccentricity, its diversity, its cheeky irreverance, its feistiness, its relentless creativity, its excellent beer, its pork pies and its pickled onions, all of which I miss, that’s why I was glad to leave it.

It saddens and maddens me to see the country of my birth killing or, at least, seriously maiming itself in such a spectacularly and wilfully ignorant way, taking the United Kingdom, and possibly even the EU itself with it, as well as causing injury to much of the world, including Canada. England is a country-sized suicide bomber. Hopefully this mob insanity will eventually be a catalyst for positive change, if not in England or Wales then at least elsewhere. Until today I opposed Scottish independence, because nationalism is almost uniformly awful and the last thing we need in the world is more separatism, but it is far better to be part of something big and expansive like the EU than an unwilling partner in something small in soul and mind like the UK. Maybe Ireland will unify and come together in Europe. Perhaps Gibraltar too. Maybe Europe, largely freed of the burden of supporting and catering for the small-minded needs of my cantankerous homeland, will rise to new heights. I hope so, but it’s a crying shame that England won’t be a part of that. 

I am proud, though, of my home city, Brighton, the place where English people who don’t want to live in England live. About 70% of Brightonians voted to stay in the EU. Today I am proudly Brightonian, proudly European, but ashamed to be English. 

 

 

Be less pigeon

Posted , updated on

I love the slogan that Audrey Watters has chosen for her new branding:

Be less pigeon

As she puts it…

“I wanted my work to both highlight the longstanding relationship between behaviorism and testing – built into the ideology and the infrastructure since ed-tech’s origins in the early twentieth century – and to remind people that there are also alternatives to treating students like animals to be trained.”

Absolutely.

Address of the bookmark: http://hackeducation.com/2016/06/08/pigeons

Can The Sims Show Us That We’re Inherently Good or Evil?

Posted

As it turns out, yes. temptations to be unkind

The good news is that we are intuitively altruistic. This doesn’t necessarily mean we are born that way. This is probably learned behaviour that co-evolves with that of those around us. The hypothesis on which this research is based (with good grounding) is that we learn through repeated interactions to behave kindly to others. At least, by far the majority of us. A few jerks (as the researchers discovered) are not intuitively generous and everyone behaves selfishly or unkindly sometimes. This is mainly because there are such jerks around, though sometimes because the perceived rewards for being a jerk might outweigh the benefits. Indeed, in almost all moral decisions, we tend to weigh benefits against harm, and it is virtually impossible to do anything at all without at least some harm being caused in some way, so the nicest of us are jerks to at least some people. It might upset the person who gave you a beautiful scarf that you wrecked it while saving a drowning child, for instance. Donating to a charity might reduce the motivation of governments to intervene in humaniarian crises. Letting a car in front of you to change lanes in front of you slows everyone in the queue behind you. Very many acts of kindness have costs to others. But, on the whole, we tend towards kindness, if only as an attitude. There is plentiful empirical evidence that this is true, some of which is referred to in the article. The researchers sought an explanation at a systemic, evolutionary level.

The researchers developed a simulation of a Prisoners’ Dilemma scenario. Traditional variants on the game make use of rational agents that weigh up defection and cooperation over time in deciding whether or not to defect, using a variety of different rules (the most effective of which is usually the simplest ‘tit-for-tat’). Their twist was to allow agents to behave ‘intuitively’ under some circumstances. Some agents were intuitively selfish, some not. In predominantly multiple round games,  “the winning agents defaulted to cooperating but deliberated if the price was right and switched to betrayal if they found they were in a one-shot game.” In predominantly one-shot games – not the norm in human societies – the always-cooperative agents died out completely. Selfish agents that deliberated did not do well in any scenario. As ever, ubiquitous selfish behaviour in a many-round game means that everyone loses, especially the selfish players.  So, wary cooperation is a winning strategy when most other people are kind, and it benefits everyone so it is a winning strategy for societies and favoured by evolution. The explanation, they suggest is that:

“when your default is to betray, the benefits of deliberating—seeing a chance to cooperate—are uncertain, depending on what your partner does. With each partner questioning the other, and each partner factoring in the partner’s questioning of oneself, the suspicion compounds until there’s zero perceived benefit to deliberating. If your default is to cooperate, however, the benefits of deliberating—occasionally acting selfishly—accrue no matter what your partner does, and therefore deliberation makes more sense.“

This accords with our natural inclinations. As Rand, one of the researchers, puts it:  “It feels good to be nice—unless the other person is a jerk. And then it feels good to be mean.” If there are no rewards for being a jerk under any circumstances, or the rewards for being kind are greater, then perhaps we can all learn to be a bit nicer.

The really good news is that, because such behaviour is learned, selfish behaviour can be modified and intuitive responses can change. In experiments, the researchers have demonstrated that this can occur within less than half an hour, albeit in a very limited and artificial single context. The researchers suggest that, in situations that reward back-stabbing and ladder-climbing (the norm in corporate culture), all it should take is a little top-down intervention such as bonuses and recognition for helpful behaviour in order to set a cultural change in motion that will ultimately become self-sustaining. I’m not totally convinced by that – extrinsic reward does not make lessons stick and the learning is lost the moment the reward is taken away. However, because cooperation is inherently better for everyone than selfishness, perhaps those that are driven by such things might realize that those extrinsic rewards they crave are far better achieved through altruism than through selfishness as long as most people are acting that way most of the time, and this might be a way to help create such a culture.  Getting rid of divisive and counter-productive extrinsic motivation, such as performance-related pay, might be a better (or at least complementary) long-term approach.

Address of the bookmark: http://nautil.us/issue/37/currents/selfishness-is-learned

This is the Teenage Brain on Social Media

Posted

An article in Neuroscience News about a recent (paywalled – grr) brain-scan study of teenagers, predictably finding that having your photos liked on social media sparks off a lot of brain activity, notably in areas associated with reward, as well as social activity and visual attention. So far so so, and a bit odd that this is what Neuroscience News chose to focus on, because that’s only a small subsection of the study and by far the least interesting part. What’s really interesting to me about the study is that the researchers mainly investigated the effects of existing likes (or, as they put it ‘quanitfiable social endorsements’) on whether teens liked a photo, and scanned their brains while doing so. As countless other studies (including mine) have suggested, not just for teens, the effects were significant. As many studies have previously shown, photos endorsed by peers – even strangers – are a great deal more likely to be liked, regardless of their content. The researchers actually faked the likes and noted that the effect was the same whether showing ‘neutral’ content or risky behaviours like smoking and drinking. Unlike most existing studies, the researchers feel confident to describe this in terms of peer-approval and conformity, thanks to the brain scans. As the abstract puts it:

“Viewing photos with many (compared with few) likes was associated with greater activity in neural regions implicated in reward processing, social cognition, imitation, and attention.”

The paper itself is a bit fuzzy about which areas are activated under which conditions: not being adept at reading brain scans, I am still unsure about whether social cognition played a similarly important role when seeing likes of one’s own photos compared with others liked by many people, though there are clearly some significant differences between the two. This bothers me a bit because, within the discussion of the study itself, they say:

“Adolescents model appropriate behavior and interests through the images they post (behavioral display) and reinforce peers’ behavior through the provision of likes (behavioral reinforcement). Unlike offline forms of peer influence, however, quantifiable social endorsement is straightforward, unambiguous, and, as the name suggests, purely quantitative.”

I don’t think this is a full explanation as it is confounded by the instrument used. An alternative plausible explanation is that, when unsure of our own judgement, we use other cues (which, in this case, can only ever come from other people thanks to the design of the system) to help make up our minds. A similar effect would have been observed using other cues such as, for example, list position or size, with no reference to how many others had liked the photos or not. Most of us (at least, most that don’t know how Google works) do not see the ordering of Google Search results as social endorsement, though that is exactly what it is, but list position is incredibly influential in our choice of links to click and, presumably, our neural responses to such items on the page. It would be interesting to further explore the extent to which the perception of value comes from the fact that it is liked by peers as opposed to the fact that the system itself (a proxy expert) is highlighting an image as important. My suspicion is that there might be a quantifiable social effect, at least in some subjects, but it might not be as large as that shown here. There’s very good evidence that subjects scanned much-like photos with greater care, which accords with other studies in the area, though it does not necessarily correlate with greater social conformity. As ever, we look for patterns and highlights to help guide our behaviours – we do not and cannot treat all data as equal.

There’s a lot of really interesting stuff in this apart from that though. I am particularly interested in the activiation of the frontal gyrus, previously associated with imitation, when looking at much liked photos. This is highly significant in the transmission of memes as well as in social learning generally.

Address of the bookmark: http://neurosciencenews.com/nucleus-accumbens-social-media-4348/

Bigotry and learning analytics

Posted , updated on

Unsurprisingly, when you use averages to make decisions about actions concerning individual people, they reinforce biases. This is exactly the basis of bigotry, racism, sexism and a host of other well-known evils, so programming such bias into analytics software is beyond a bad idea. This article describes how algorithmic systems are used to help make decisions about things like bail and sentencing in courts. Though race is not explicitly taken into account, correlates like poverty and acquaintance with people that have police records are included. In a perfectly vicious circle, the system reinforces biases over time. To make matters worse, this particular system uses secret algorithms, so there is no accountability and not much of a feedback loop to improve them if they are in error.

This matters to educators because this is very similar to what much learning analytics does too (there are exceptions, especially when used solely for research purposes). It looks at past activity, however that is measured, compares it to more or less discriminatory averages or similar aggregates of other learners’ past activity, and then attempts to guide future behaviour of individuals (teachers or students) based on the differences. This latter step is where things can go badly wrong, but there would be little point in doing it otherwise. The better examples inform rather than adapt, allowing a human intermediary to make decisions, but that’s exactly what the algorithmic risk assessment described in the article does too and it is just as risky. The worst examples attempt to directly guide learners, sometimes adapting content to suit their perceived needs. This is a terribly dangerous idea.

Address of the bookmark: http://boingboing.net/2016/05/24/algorithmic-risk-assessment-h.html

A blueprint for breakthroughs: Federally funded education research in 2016 and beyond | Christensen Institute

Posted

An interesting proposal from Horn & Fisher that fills in one of the most gaping holes in conventional quantitative research in education (specifically randomized controlled trials but also less rigorous efforts like A/B testing etc) by explicitly looking at the differences in those that do not fit in the average curve – the ones that do not benefit, or that benefit to an unusual degree, the outliers. As the authors say:

“… the ability to predict what works, for which students, in what circumstances, will be crucial for building effective, personalized-learning environments. The current education research paradigm, however, stops short of offering this predictive power and gets stuck measuring average student and sub-group outcomes and drawing conclusions based on correlations, with little insight into the discrete, particular contexts and causal factors that yield student success or failure. Those observations that do move toward a causal understanding often stop short of helping understand why a given intervention or methodology works in certain circumstances, but not in others.“

I have mixed feelings about this. Yes, this process of iterative refinement is a much better idea than simply looking at improvements in averages (with no clear causal links) and they are entirely right to critique those that use such methods but:

a) I don’t think it will ever succeed in the way it hopes, because every context is significantly different and this is a complex design problem, where even miniscule differences can have huge effects. Learning never repeats twice. Though much improved on what it replaces, it is still trying to make sense through tools of reductive materialism whereas what we are dealing with, and what the authors’ critique implies, is a different kind of problem. Seeking this kind of answer is like seeking the formula for painting a masterpiece. It’s only ever partially (at best) about methodologies and techniques, and it is always possible to invent new ones that change everything.

b) It relies on the assumption that we know exactly what we are looking for: that what we seek to measure is the thing that matters. It might be exactly what is needed for personalized education (where you find better ways to make students behave the way you want them to behave) but exactly the opposite for personal education (where every case is different, where education is seen as changing the whole person in unfathomably rich and complex ways).

That said, I welcome any attempts to stop the absurdity of trying to intervene in ways that benefit the (virtually non-existent) average student and that instead attempt to focus on each student. This is a step in the right direction.

 

augmented research cycle

Address of the bookmark: http://www.christenseninstitute.org/publications/a-blueprint-for-breakthroughs/