7 min read

Reinvent Invention

Reinvent Invention

Invention, discovery, innovation, collective intelligence…we have so much more to learn about how to explore the unknown. In this week’s selection of recent research we’ll explore the very mystery of the mysterious.

Invent what you know

A couple of recent papers have used NLP (AI for text analysis) to qualify innovation by women. One paper "read" 1.2 million US doctoral dissertations and found that women in male-dominated fields produced greater innovation even as they had less successful careers and were less cited by peers…

The Diversity–Innovation Paradox in Science

Another paper "read" all biotech patents and found that all-female teams were 35% more likely to invent biotech for women's health, but...

Who do we invent for? Patents by women focus more on women's health, but few women get to invent

...just as in academia, "...men...are less likely to adopt women’s ideas" in biotech...

Marginalized and Overlooked? Minoritized Groups and the Adoption of New Scientific Ideas

I love the creative use of NLP to explore the nature of innovation. I hate that humanity cannot get out of its own way and see value in the ideas of others.

Disagree with yourself

The "Wisdom of Crowds" isn't always so wise, but there are clear benefits under the right circumstances, one of which is to keep it "fresh".

People that haven't collaborated before produce ideas of greater originality and broader impact…

Fresh teams are associated with original and multidisciplinary research

But what if the crowd is just you? Then try disagreeing with yourself.

After developing your own ideas, imagine someone with whom you frequently disagree. What would they think? Integrating those two together produces "highly accurate inner crowds"…

Taking a Disagreeing Perspective Improves the Accuracy of People’s Quantitative Estimates

And if all else fails, make your own guesses more independent by approaching a problem in distinct cognitive or even emotional contexts. This increases the "freshness" of your internal ideation…

Unfortunately, having a prodigious working memory (usually a good thing) actually hurts here. Guesstimates over time are more correlated for high-WM people. Perhaps a little beer and codeine to improve your inner crowd...while slowly ruining your life.

Smaller is better (when sampling from the crowd within): Low memory span individuals benefit more from multiple opportunities for estimation

Invent messy!

Some researchers ignored a classical solution in echolocation & opted to invent a messy solution that worked even better!

They built "a soft-robotic sensor that mimics fast non-rigid deformation of the ears in certain bat species" & then…

Integration of deep learning and soft robotics for a biomimetic approach to nonlinear sensing

…they developed a DNN that learned the "nonlinear Doppler shift signatures generated by these motions to estimate the direction of a sound source..."

Too cool! Invent messy!

Fast information, slow innovation

In "Distributed Innovation" I made a big claim: beyond a certain point, as the speed and density of information increases, innovation decreases.

Some were skeptical. Now new research finds exactly this phenomenon in science: "When the number of papers published per year in a scientific field grows large,

  1. citations flow disproportionately to already well-cited papers
  2. the list of most-cited papers ossifies
  3. "new papers are unlikely to ever become highly cited".
  4. "when they do, it is not through a gradual, cumulative process of attention gathering"
  5. "newly published papers become unlikely to disrupt existing work"…

Even the title of the paper is ominous: "Slowed canonical progress in large fields of science".

Knowledge is information; innovation is practice.

Collective Intelligence: the cognitive, creative capacity of a group.

Here are 10 important findings…

1. Equitable gender representation (or even a slight female bias) improves collective intelligence, as does other forms of complementary diversity.

2. Collective Intelligence increases with the average social perceptiveness of group members

Very much in line with research on #PsychologicalSafety, collective intelligence is maximized when all members feel free to take risks and explore.

3. Collective Intelligence generalizes well to new problems and "predicts performance on various out-of-sample tasks". It's more than just a set of specific skills.

4. "Group collaboration process is more important…than the skill of individual members" for Collective Intelligence.

In a labor market battling over the most elite individuals, it's crucial to focus instead on the quality of teams and how they collaborate. In fact…

5. In complex tasks, "interacting groups are as fast as the fastest individual and more efficient than the most efficient individual"


6. "Interacting groups generate more solutions more rapidly and explore the solution space more broadly than independent problem solvers"

Which shouldn't surprise us given that…

7. "People estimated that they generated 75% of the solution space when in fact their ideas covered only 20–30%"

Teams with complimentary diversity discover more unique solutions, increasing Collective Intelligence.

8. "Aggregating opinions of modular crowds composed of small independent groups achieved better forecasts than aggregating a similar number of forecasts from non-modular ones."

9. Some simple tests correlate with creativity, such as "name 10 words that are as different from each other as possible." Unsurprisingly, collective intelligence out performs individuals on tasks like these.

Finally, reinforcing the importance of collaboration process and social perceptiveness in #CollectiveIntelligence:

10. In biotech, "...men...are less likely to adopt women’s ideas" as well as those of all outgroups.

Creativity is Effortful

Creativity is often celebrated as a history of “Aha!” moments, but in truth, creativity is effortful.

A recent review explores this “insight bias”, in which we “associate creativity with effortless insight and undervalue persistence”.

In creative problem solving tasks “people consistently underestimate how many [additional] ideas they would generate” with further effort and overestimate how much of the problem space they have actually explored.

For example, “in one study, people estimated that they generated 75% of the [possible solutions to a problem] when in fact their ideas covered only

In another study, “people…preferred an innately talented entrepreneur with fewer achievements over a hard-working entrepreneur with more achievements.”

When I was young, I thought that if something wasn’t easy then I must not be good at it.

And so I gave up on everything.

Now I’m older and nothing is ever easy. I’m not good at anything.

…but I keep trying.

And every so often I create something extraordinary.

Though there is much to be said for dumb luck and good-hearted theft.

Raʼs al Ghul was right

The premature death of an eminent...scientist” decreases articles by “collaborators in affected fields,” but “the flow of articles by non-collaborators increases markedly,” draws “upon a different scientific corpus”, and is much more “likely to be highly cited.”

Furthermore, over the last few decades “while a larger proportion” of scientific papers have been “cited at least a few times, citations are also more concentrated at the top of the citation distribution.” Gini coefficients raise and winners-take-all (or close to it).

Should we follow the sage advice of “Logan’s Run” and institute Carousel for all scientists and inventors over 30? (And musicians and directors for that matter?) Or do we acknowledge that we all herd around safe ideas and forced ourselves to change?

True innovation must surprise

Science papers that “make more (distant) new combinations...are more likely” to

  • “be a top 1% highly cited paper”,
  • “inspire follow-on highly cited research”
  • “be cited in a broader set of disciplines”

But initially they find little love

Recognition for novel papers is slow coming. They are

  • “less likely to be top cited” in their first few years of publication,
  • “significantly more highly cited” outside “their ‘home’ field”, and
  • “published in journals with a lower Impact Factor”

Perhaps the most important lesson for anyone hoping to innovate is that greater novelty means “higher variance in its citation performance”. In other words, failure rates are higher. Innovation and uncertainty travel together through the unknown.

Market games game markets

There’s an idea that intense, high-stakes competition drives elite performance (see “Whiplash”), but in an experimental art market game, "competition does not significantly increase…creativity".

The intense competition "also leads to more unfair reviews and to a lower level of agreement between reviewers." And the “higher rejection rate under competitive conditions does not improve the average quality of published contributions, because more high-quality work is also rejected."

Related research found that "when many contestants compete for a few [rewards], strategic contestants adopt high-risk strategies" that "reduce the correlation between performance and ability."

It may be that as in the real world, the participants in the art competition were exploiting strategies to maximize rewards in ways that do not increase the desired outcome: creativity.

I wonder how this finding interacts with the research on innovation and "minority opinion" (e.g., you only receive rewards for ideas that the majority fails to discover). How do assessment rules (e.g., peer review, majority rule, unanimous opinion, etc.) interact with reward structure (e.g., market, minority opinion, individual reward, etc.)?

The B-Flats

In an analysis of thousands of collaborations, I found the small, flat, diverse teams produce the most innovation (this will be a new upcoming chapter).

A new paper in PNAS puts an exclamation point on one of those factors: flatness.

By measuring the “ratio of members playing leadership roles to total team size”, the research shows (across 16,397,750 papers) that “relative to flat, egalitarian teams, tall, hierarchical teams”

  1. “produce less novelty and more often develop existing ideas”
  2. “increase productivity for those on top and decrease it for those beneath”
  3. “increase short-term citations but decrease long-term influence.”

In fact, “the same person on the same-sized team produces science much more likely to disruptively innovate if they work on a flat, egalitarian team”.

Many tasks may require decisive bosses; innovation requires complimentary collaborators.

"Diversify!" said the AI.

A computational model designed to robustly predict replicability scientific claims found that “scientifically focused but socially and institutionally diverse research activity is most likely to replicate.”

The model argues for “sponsorship of increased diversity of and independence between investigations of any particular scientific phenomenon, and diversity of scientific phenomena investigated.”

In other words, NIH, stop biasing funding in favor of safe ideas from long established labs. We cannot discover something new only by exploring what we already know. Innovation requires testing intriguing ideas that might not work but that will fundamentally change our understanding if they do. It requires failure.

The Long Goodbye

I was born in LA in 1971, the setting of Robert Altman's "The Long Goodbye". Watching Elliot Gould snark his way through the plot, I was struck by how little my world has changed over my lifetime. Ignoring the occasional rotary phone or bell-bottom, Gould's day feels like it could fit right into 2021 LA...

50 years and so little fundamental change in my world. There have been plenty of incremental improvements under the hood, but 50 years before my birth, 1921, America would be a largely alien world to me. Is this stagnation or stability?

BTW - Cinematically, I recommend the Altman's remake. It satires the twisty grit of Hawks' "The Big Sleep" (1946), with ineffectual Gould replacing hard-boiled Humphrey Bogart. He is Philip Marlowe the way Dick Shawn is Hitler.