How do you set up fertile ground for breakthroughs to emerge? What makes extraordinary teams persist? How do their ideas go from incubation to execution?
For the past two months, I have been part of a great group discussing everything revolving around the Great Groups of the past, in the form of a book club set up by Arnaud Schenk and Matt Clifford. From the Manhattan Project to the American Revolution and everything in between, we tried to analyze common and divergent threads that ran through some incredible efforts of the past.
I spent much time after each session elaborating on points, summarizing my thoughts, and posing additional questions. So why not share those with others who might be interested in these topics as well?
What follows is the first part of this mini-series. Although some assumptions made in the summaries are the product of the group’s conversations and thus aren’t built from the ground up for those who did not attend, some thoughts and pointers might lead readers to further inquiry and down the rabbit hole.
So on we go!
Part 1: Organizing Genius – The Great Technology Labs:
What are the best conditions for great ideas to take shape? How do they go from incubation to execution? What can we, in the 21st century, learn from the great technology labs of the 1900s? From looking at Bell Labs, Skunk Works, Xerox PARC, and ARPA, a few common themes emerge, whilst other aspects of their structures differ immensely. And for every potential answer, an array of new questions can be asked.
Nonetheless, in the spirit of cross-fertilization of perhaps unrelated ideas, here’s a pondering of some of the themes of Organizing Geniuses fourth session on the Great Technology Labs.
How much innovation can you plan?
While discussing the characteristics of the labs, a recurrent theme was the dichotomy between bottom-up and top-down approaches to innovation.
What all labs seemed to have in common was their serendipitous flow of ideas funneled by the interdisciplinarity of their teams. Water-cooler-conversations between two curious researchers could result in advances we still use today – an afternoon’s invention of overlapping windows after a round of beer being such an example at PARC. The great labs overall seem to have had an organic way of innovating.
An interesting aspect of these groups, whilst from different backgrounds, was their relative homology once they were in the same space. While taking different approaches, they generally believed in the same vision and had a cultural identity around it. This notably enabled the formation of trust which in turn was crucial to a truly free flow of ideas. Oftentimes friends were reeled into the organizations, which resulted in a way of recruiting not very unlike nepotism, but possibly essential to the existence of trust. It’s interesting to think about whether this characteristic was due to the times or in fact more fundamental to the revolutionary discoveries than we would perhaps like to acknowledge. Nonetheless, there may be other ways to ensure a more open process of recruiting great teams and further research into this question seems quite important.
In contrast, we addressed some more top-down attempts to fuel progress, like the UK’s tax credits on R&D, oftentimes resulting in misaligned incentives of governments and corporations. Heavily state-orchestrated innovation seems like an oxymoron after looking at the successful labs.
While we can never factor out the crucial elements of serendipity, as well as the aligned external circumstances in markets and politics that make certain labs radically successful, we can probably take away some vital fundamentals based on the common threads of the past examples. There seem to be things that are certainly more likely to cause the emergence of the hard-to-quantify qualities of innovation. On the team-basis, the following were especially talked about:
- interdisciplinary exchange of ideas -> requires people from different backgrounds, encouragement of dialogue from management, and environment in which they can bump into one another (i.e. physical spaces much more likely to foster this exchange)
- great managers of people (e.g. Oppenheimer)
- environment of trust (cultural homology -> belief in the same vision)
Peter Thiel has in the past formulated “Thiel’s Law” which says: “a startup messed up at its foundation cannot be fixed“. The same seems to apply to research labs and more generally to the formation of any team.
Research with a clear goal vs open-ended inquiry
David Hackett-Fischer once asserted that “questions are the engines of intellect, the cerebral machines which convert energy to motion, and curiosity to controlled inquiry”.
A challenge that any research project faces is the balance between open-ended and bounded exploration. While Skunk Works had one target customer (the military) and pretty defined goalposts of what their designed planes had to be able to do, Xerox adopted the goal of creating “the office of the future”, a much more broadly definable statement, with no immediate customers to cater to (since Xerox itself was averse to directly commercialize many of the inventions).
This proved to be detrimental in the eyes of some researchers, like Alan Kay: “The worst thing that Xerox ever did was to describe something as the office of the future, because if something is the office of the future, you never finish it.” This is somewhat reminiscent of the currently relevant Tesler’s Theorem which asserts that “AI is whatever hasn’t been done yet”.
The nature of the lab’s vision greatly matters in shaping the collective focus of the group and thus getting stuff done (and importantly, making the members feel like they are getting stuff done by e.g. getting products out there/ hitting official goalposts/ etc).
It’s obviously hard to determine what the ideal difficulty of the goal is in advance. By definition, basic research can’t be predetermined in this manner.
Importantly, research labs often differ in how nascent their ideas are. Skunk Works built planes that were known to be buildable. ARPA funds projects partly determined by expert opinions on their executability.
The most productive goals of the great labs seemed to be hard, but doable and ideally also addressed a tangible need, either of the customer or of the researchers themselves (e.g. Intel chip testers being built by PARC researchers to advance faster on their other projects). This tangible need brings us to the next point.
Parent organizations & their research labs – aligning incentives
The researchers’ wish of impact probably shouldn’t be underrated either. Apart from of the parent organization, at the end of the day it is also in their interest to get innovations out there in the form of products. At PARC, the mismatch between Xerox’s ambiguous goals of commercialization (or more often than not the absence thereof) and the researchers’ wish to see their inventions in their friends’ and families’ homes was made clear by various examples of employees taking matters into their own hands by either leaving to start their own companies or lobbying the Xerox executive rank to bring the work to market, as in Larry Tesler’s case (concerning the NoteTaker). In cases where there is no direct customer and no set deadline, it seems important for a parent organization to communicate plans with its research lab and to make sure that incentives are aligned. Researchers should know the purpose of their being there and the prospects of their inventions, shall they be successful. For more open-ended research examples like PARC, the question of exactly when to take the leap from project to commercialization remains. This question of course can’t be separated from the individual circumstances at the time and answers will thus likely vary case by case.
A new funding landscape & deep tech progress going forward
While trying to draw conclusions from the great technology labs of the past, we also addressed the present and future of deep tech progress. As always, funding plays a huge role and thus it is useful to look into how different financing models could aid further technological advances. We generally agreed that VC oftentimes has too short of a time horizon for much fundamental research to take place. Projects that have a 7+ year timespan to (potential) commercialization aren’t attractive to investors whose payday arrives through their not-too-far-out exit. But we also saw how ARPA PMs have a tight timeframe of 3-5 years to complete their ambitious projects and noted the risks of “patient money” in creating unfavorable incentives that may align against the focus and targeted effort which shorter deadlines ensure. Again, it depends on the nature of the project, but great progress has historically been shown to be achievable in surprisingly short amounts of time. Maybe an important aspect in getting investors on board for deep tech advances is structuring the projects in a way that ensures that goalposts can be met in segments while as a result also keeping the team on track to gradually getting to the peak of the mountain.
Another issue concerns the right way of government funding. What makes the US radically more innovative than Europe? Is this difference more influenced by the amount or the mechanisms of public funding? What could Europe learn from the foreign examples?
Lenient quotas for R&D tax credit which can be filled through unrelated areas like HR don’t seem like a promising way to go. Perhaps it is also a cultural vision of optimism vs pessimism that people like Thiel have advanced throughout the years that makes the US significantly more innovative and wealthy as a result.
In addition to non-optimal funding mechanisms, we also don’t see the past century’s drive for essential progress in deep tech. Instead, a lot of talent goes into the very lightly regulated field of software which -while certainly an area whose potential is nowhere near exhausted- is nowadays often driven by incremental progress. This field presents a comparably low barrier of entry in terms of degrees (many self-taught software engineers and entrepreneurs) and little upfront cost to get started and even commercialize later on. The same cannot be said about deep tech which is known for heavy regulation and expected credentials to start ventures, based on specific expertise required for many advancements.
As we move forward in this century, the question of how much self-selection out of the potential deep tech talent pool is caused by regulation and credentialism becomes increasingly important to investigate.
Readings for this session included chapters of The Idea Factory by John Gertner, Organizing Genius by Patricia Biedermann and Warren Berger, and the article Inside the PARC.