A private ARPA requires new legal organization structures

Exploratory Program Organization

Academia is not a good environment for systems engineering

Academia incentivizes novelty, not focus

Venture capital doesn’t work for things that only realize their value on long timescales

Academic culture prizes individual recognition

All funding organizations are research organizations

We have no flying cars - failures of paradigm-shifting engineering

Things that are not paper-worthy enough for academia and not product-focused enough for a startup

In the past, corporate labs filled a particular niche in the innovation ecosystem

Academia judges via novelty because there are no metrics but there are no metrics when the thing is still research

Ilan Gur Conversation 3 Sep 2020

# A new structure for scalable research
` The vast majority of new ideas and technologies are never reproduced or achieve large-scale impact`
- [[The vast majority of new ideas and technologies are never reproduced or achieve large-scale impact]]
`In part, this is because Academic culture prizes individual recognition`
- [[Academic culture prizes individual recognition]]
- Things that `Only realize their value on timescales beyond the horizons of typical venture investment, or that depend for their impact on results being public`::Need list of examples of these::
- [[Venture capital doesn’t work for things that only realize their value on long timescales]]
- [[There is a significant class of innovations that would create drastically less value for the world if their value had been captured by their creators]]
- `Academia incentivizes novelty, not focus`
- [[Academia incentivizes novelty, not focus]]
- `Innovation is one of the key criteria by which grant applications and academic publications are judged. To fulfill the innovation criterion, academics are incentivized to distinguish themselves from others and to pursue novel research. In the academic sciences, novelty comes in the form of new discoveries, whereas in academic engineering it comes in the form of proofs of concept. `
- [[Academia judges via novelty because there are no metrics but there are no metrics when the thing is still research]]
- Incentive structure leads to limitations on 
	1. Accountability 
	2. Transparency 
	3. Collaboration
- [[The academic incentive structure leads to limitations on accountability, transparency, and collaboration]]
- `Academia occupies a critical point in the innovation cycle, as the primary source of new ideas ` ::Not sure I agree, but it’s definitely *a* source of new ideas::
- FRO principle investigators = PARPA program managers
- `Metrics driven`::need to have a good way of establishing the right metrics::
- No secrets except to protect IP ::this is a tricky piece::
- `Janelia has found that “without an opposing force provided by management, there is a slow, steady drift toward a more conventional environment increasingly focused on maintaining successful programs and documenting individual achievement at the expense of risk taking and collaborative, interdisciplinary work” `
- [[Deliverables vs metrics]]
- [[Startups  need to spend effort on sales, marketing, and productization]]
- “Low profit LLCs”
- `Alternatively, one could imagine the creation of a dedicated umbrella organization or a government program with the specific goal of initiating and supporting FROs. `
- Some interesting numbers on how much value MIT would need to capture 
- On board with 3-4 years and then terminating
	- 
- `Each FRO should have its own board of directors` This seems like massive overheard 
### Published
2019	
## Questions
## Thoughts
- We need to talk about low hanging fruit of existing systems. Insanity is doing the same thing and expecting different results
	- [[Low hanging fruit narratives don’t take context into account]]
	- Misses the overhead of what it would take to spin up each FRO, getting the people to join etc

## Related
Draft roadmap for second phase of BRAIN initiative 
- [[Focused Research Organization]]
- 

### Cites 
	- A. P. Alivisatos et al., A National Network of Neurotechnology Centers for the BRAIN  Initiative. Neuron. 88, 445–448 (2015). ::roadmapping example?::
		- https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5283530/pdf/nihms844886.pdf
	- The Comptroller General of the United States, “[[Nanomanufacturing -  Emergence and Implications for U.S. Competitiveness, the Environment, and Human Health]]” (2014), (available at https://www.gao.gov/assets/670/660591.pdf). 
	- G. M. Rubin, E. K. O. Shea, Looking back and looking forward at Janelia (2006). ::incentive drift::
	- M. Watzinger, T. A. Fackler, M. Nagler, M. Schnitzer, How Antitrust Enforcement Can Spur Innovation: Bell Labs and the 1956 Consent Decree (2017) (available at https://ssrn.com/abstract=2904315). 
	-  D. W. Braben, Pioneering research: A risk worth taking (John Wiley & Sons, Ltd, 2004). 
	- J. Segal, How Social Entrepreneurship is Changing Chicago (and the World). Technori  (2012). ::low profit llc::

## Authors
- [[Sam Rodriques]]
## Link
https://www.dropbox.com/s/piswunee00nfjn4/A%20New%20Structure%20for%20Scalable%20Research.pdf?dl=0

[Web URL for this note](http://notes.benjaminreinhardt.com/A_new_structure_for_scalable_research)

[Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/A_new_structure_for_scalable_research)






§Corporate R+D Constraints

DARPA’s aversion to people with a web presence may be how they avoid asymmetric career risk

One of the most failure-prone points in technological development is when a technology needs to move from an organization that focuses on R&D to one that focuses on marketing and manufacturing

# Asymmetric career risk impacts corporate R&D
Someone always needs to go out of their way to make the output of a corporate R&D lab mainstream because  [[R&D orgs are always on the side of an org chart - The lab is always separated from the mothership]] and [[There is rarely a default path for R&D output to become mainstream]]. [[Asymmetric career risk]] is especially acute for this person because it is never ‘their job’ to push the technology through.  The solution to depending on asymmetric career risk for new things to ram their way into a company is to [[Manage the Transfer not the Technology]], but setting that up in the first place is going against the stream. It’s just easier to have [[Innovation Theater]].

In the late 2010’s companies tried to address this problem by creating “innovation teams.” From conversations with people at these arms - there is still no default bridge crosser.

If you look at the heralded labs of the 20th century like [[Lockheed Skunkworks]] and the [[Rad Lab]] many of them came within a hair of failure and only succeeded because of a champion. 

This is also a problem for startups trying to sell into large companies. 
[[Asymmetric career risk]] happens across all levels of an organization. CEOs are fired by boards for trying weird things as often as engineers are fired by managers for not putting their heads down and working on what they’re supposed to. 

### References
-  [[Conversation with Kelly 12/2]] 



[Web URL for this note](http://notes.benjaminreinhardt.com/Asymmetric_career_risk_impacts_corporate_R-D)

[Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/Asymmetric_career_risk_impacts_corporate_R-D)




Commercializing non-health research does not produce VC level returns

Isaac Barchas Conversation 18/3/20

# Commercialization has very high transaction costs
- Finding the people who have the right criteria is hard. They need to 
	- have done research that’s at the correct TRL 
	- want to do the work to spin it out 
	- be in a position to spin it out (ie. Tech transfer office is friendly)
- Away from the coasts  there are fewer or zero lawyers and accountants who are used to working with startups
	- This makes it so there less network lubricant for due diligence so it has higher transactions costs. [[The best due diligence is network based]] 
	-  Fewer or zero infrastructure people 
- In places where it’s not in the culture people just don’t know what the fuck they’re doing around commercialization 

### Thoughts
- Could something like [[Polyplexus]] be used to lower transaction costs?



[Web URL for this note](http://notes.benjaminreinhardt.com/Commercialization_has_very_high_transaction_costs)

[Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/Commercialization_has_very_high_transaction_costs)




When is push or pull appropriate for innovation orgs?

Ideally corporate R+D enables targeted dicking around

Bell Labs hired people before they had a role or expertise

ICT Distillation

An Institute for Creative Technology - Riffing on ARPA

ICT Outline

# Comparing and contrasting Bell labs, DARPA, and Venture Research
Venture Research, Bell Labs, and DARPA are all successful innovation organizational paradigms that feel almost like a set of *non-orthogonal* basis vectors. [[Venture Research, Bell Labs, and DARPA form non-orthogonal basis vectors over the space of innovation organizations]]

High level, Venture Research is squarely aimed at [[Einstein would have been stuck in a patent office - failures of paradigm-shifting science]] while Bell Labs was targeted at [[We have no flying cars - failures of paradigm-shifting engineering]]. 

**The organizations differed in how they recruited people.** Specifically how much each cares about skills, idea crispness, and self-motivation. These different requirements in turn manifest in actual recruiting tactics. Venture research requires all three -  idea crispness, self-motivation, and skills. Venture research puts the least weight on skills, with the assumption that the idea’s precision indicates skill or the ability to find people to have it. The coupling between idea precision and skill to execute on it feels intuitively true, but this is an intuition I’m a bit skeptical of.  Perhaps it depends on the scope of the project. 

Venture research recruited people by strategically making groups of potential applicants aware of its existence, but then waits for applicants to apply. [[Howard Hughes Medical Institute (HHMI)]] [does the same thing.](https://www.hhmi.org/programs/biomedical-research/investigator-program#Application)  The justification (that makes sense) is that the ideas are going to be so out there and the researcher must be so obsessed with them that applying is part of the test. Additionally, they had the hypothesis that the ideas could come from relatively unknown researchers and solicitations would inherently be biased towards more well-known researchers. 

[[Bell Labs hired people before they had a role or expertise]] - “hey you’re smart and know about electrons, you should come work here.” That is, high focus on general skill but not specific skill and people could have zero idea crispness. In-house organizations have the luxury of ramping up to self-motivation. 

DARPA falls somewhere in between. PMs need to be extremely self-motivated day one, but their direction can be internal or external depending on where the idea inception comes from.  [[Ideas can be incepted from many places]]. The first part of a PMs tenure is focused on sharpening ideas so you don’t need crisp ideas day one. Skills is the tricky one here because while DARPA PMs aren’t the ones actually executing (unlike Venture Research or a Lab) they do need to both be able to quickly create a precise plan and earn the respect of performers (and call bullshit on them.)  Since [[People who have done a thing should be in charge of a thing]], that suggests that they do need skill beyond basic knowledge. DARPA is very much a “you don’t call us, we call you” system. 

**Idea Inception**  [[Where do ideas come from?]]
DARPA has a spectrum of idea inception. In the “idealized” version of DARPA the inception of the vision at least comes from the PM. In reality, the idea can either come from above (the Director/DoD say “we need this”) or below, via the PM talking to experts in a general area and being suggested a specific program idea. In reality the inception is probably an informal iterated mix of the the three (Director has a hunch, hires a PM who has a notion similar to the hunch, they talk about it, the PM holds a workshop where she throws the notion around with some experts and in her head it becomes an idea.) 

Venture research requires that the idea be incepted completely by the performer. The venture research team didn’t even solicit proposals from people they thought might be good, which could have ‘corrupted’ the idea inception by implicitly saying “hey we think the things you’ve done in the past might be worthwhile to build off of.” 

Bell Labs seems to have had a mix of idea inception. Sometimes it was someone at the top saying “look at this.” Sometimes it was someone like Shannon being very interested in something. The size of the organization supported that mix. It was also hierarchical, in that the “box” inception would be external (“figure out how to make pure germanium”) but within that box there was a lot of inceptive leeway. 

Both the different approaches to idea inception and recruiting abstract to what I would call “push vs. pull” approaches. Where to fall on the push-pull spectrum is an important thing for new innovation organizations to think about. [[When is push or pull appropriate for innovation orgs?]].

**The organizations differed in where the locus of control and risk is located.**  [[DARPA Program managers pull control and risk away from both researchers and directors]].  There’s a spectrum for DARPA on the level of directional control PMs have. Sometimes DARPA will hire program managers because they are smart and are working in an area that’s vaguely interesting to the DoD’s interests. On the other hand, sometimes DARPA will hire program managers to work on a *specific program* like Have Blue. Notably, the first stealth fighter was built in conjunction with [[Lockheed Skunkworks]] who both institutionally resembled [[Bell Labs]] and were the instigators for the design - so arguably they took on both more risk and more control. [^1][^2]

Venture research puts *all* of the control and risk in the hands of the researchers, once they have the go-ahead. 

Bell Labs put a lot of control in the hands of individuals and research teams. It also seems to have pushed the decision making process as far down as possible - see the anecdote about not showing (or needing to show) demos to higher management until they were sure they had a thing. [[Bell Labs management was extremely light]]. However, the management would suggest high level things that people should work on (“Why don’t you take a look at this mobile phones memo”) and would cut off projects (see Pierce’s rules about killing projects without damning the researcher.)  That latter point about not damning the researcher brings up a question about risk - who *does* bear the risk? There’s also the fact that researchers seem to have a lot of control *within* the constraint of Bell Labs’ mandate to do work that will eventually improve the AT&T system. This “control box” can also be seen at DARPA where PMs have a certain amount of money to play with and need to work on things that the DoD will find relevant. 

**The organizations have different feedback loops** 
[[Feedback loops matter even if there is no control explicitly being exerted]]. 

At DARPA, the PM’s feedback loop with the Director is based on clear tranches. The PMs have a tight feedback loop with the performers - in some cases, the teams touch base every week and can fire them at a moment’s notice. Of course, that also means that the PMs have a ton of context and can provide supplemental funding and other help at a moment’s notice as well. Such a tight feedback loop requires two pieces that are easy to leave behind. First, there needs to be a ton of trust between the PM and the performers. Second, the projects need extremely precise goals so that everybody is on the same page on whether the project 

Venture research is at the opposite end of the feedback spectrum from DARPA.  Once funding commences, the research is completely open-loop (if the researcher wants it to be) for the full five years of the grant. The open-loop model in venture research gives the researcher an incredible amount of freedom but also requires extremely high trust on the part of the funder.  [[Venture research built trust by iterated meetings to refine precision around the idea]].

 Bell Labs had more frequent and less formal loops, which they could get away with because everything was in house. [[Bell Labs management was extremely light]]. It’s not recorded anywhere but my hunch is that there were a lot of casual “how’s it going” chats but few formal status reports.

### Patterns
The three different organizational models have very different constraints and expected outputs so most of the lessons come from the deltas between the organizations. However, it’s striking that there are a few constant patterns.  

**Trust is essential for all three approaches**
[[Venture research built trust by iterated meetings to refine precision around the idea]] - in some cases, it could take up to a year for the researchers to trust the VR team and for the VR team to trust the researchers. This trust was essential because by definition it is basically impossible for anybody who didn’t have the idea to know the underlying risk distribution for paradigm-shifting work. Therefore, rather than just labelling the work ‘high risk’ and expecting it to fail, the VR team built up the trust to understand the researcher’s assessment of their probability distribution. [[Uncertainty always involves risk but risk does not always involve uncertainty]]. This trust enabled the researchers to work without worrying that constraints would appear all of a sudden and for the VR org to commit to them for five years. 

DARPA requires trust for the opposite reason VR does. Tight feedback loops require trust just as much as loose ones because the performers need to trust that they can be open with how things are actually going to the PM. Additionally in order for the tight feedback loops to exist at all, the director needs to trust PMs enough to give them free rein over spending. 

Bell Labs needed trust so that it could switch between these two modes. 

**Precision is essential for all three approaches**
[[Venture research built trust by iterated meetings to refine precision around the idea]]. Braben’s assertion that people can develop very precise ideas around paradigm-shifting research up front is counterintuitive but important. 

Venture research’s trust-through-precision-refinement could be seen as an externalized version of DARPA’s seedling programs where PM’s build trust and precision through seedling programs once they’re part of the organizations. 

### Related
- [[Bell Labs was only profitable in conjunction with AT+T]]
- [[brabenScientificFreedomElixir2008]]

[^1]:The anecdote of a powerful PM with high level direction coming from above and low-level direction coming from below, suggests a  that there are different roles in the process that *need* to be filled, but you can shift them between collaborators, if this fact is understood and transaction costs are low enough. [[What are the different roles in an innovation org?]]
[^2]: It’s also important to think about locus of control vs. perceived locus of control. Specifically, you have [[Alan Kay]]’s account of having complete free rein, while at the same time it was clearly [[JCR Licklider]]’s project, inception, and vision. There are two explanations I can see for this. One is that someone is wrong: Alan is idealizing/misremembering or Licklider had no idea what he was doing and it all just happened to come together. The other explanation is that Licklider managed to construct programmatic blocks and found people to fill them whose interests were such that they didn’t even notice the boundaries of the box because that is exactly where they wanted to be. This seems very hard. 



[Web URL for this note](http://notes.benjaminreinhardt.com/Comparing_and_contrasting_Bell_labs,_DARPA,_and_Venture_Research)

[Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/Comparing_and_contrasting_Bell_labs,_DARPA,_and_Venture_Research)




Overlapping institutional constraints rule out several classes of creative work

Research requires intrinsic motivation

# Contemporary research has multiple failure modes
When we talk about [[Stagnation]] in “research” we are lamenting some admixture of several  failure modes.[^1] 

One class of failures is in the failure to enable paradigm shifts. You can roughly divide the failure to enable paradigm shifts into two modes. The first mode is the “Einstein would be stuck in the patent office” mode:  we don’t seem to have a healthy system of unfettered research that is producing paradigm shifts like those seen up through the 50’s or 60’s. The second mode is the “Where’s my flying car?” Mode - we don’t seem to be able to build new physical systems that permeate into the world. One might roughly call these two “paradigm-shifting science” and “paradigm-shifting engineering” but the line between science and engineering is [[Nebulous]], porous, and full of feedback loops [[Research has many orthogonal and non orthogonal classification axes]]. 

Interestingly, [[odlyzkoDeclineUnfetteredResearch1995]] ties together both [[Einstein would have been stuck in a patent office - failures of paradigm-shifting science]] and  [[We have no flying cars - failures of paradigm-shifting engineering]]. 

Another class of failures is what I might call “breakdowns in the scientific process.” Here you see issues like the replication crisis and science being judged not on one of several scientific epistemologies, but on politics and its ilk.

 It’s important to call out these different modes because while they share many similarities and causes, precision in language leads to precise thoughts [[Precise Language Enables Progress]]  and the steps to address each mode are likely different. It’s also important to example the causal links between them, which might not be as tight as we think despite sharing confounding variables.  

### Related
- [[It is important to precisely define problems in an area]]
- [[Precise communication enables other people to make decisions]]
- [[Phenomena-based cycles]]
- Here I am being a splitter [[Lumpers and Splitters]]

[^1]: Papers like [[collisonScienceGettingLess2018]] lump “research”  into one category and so does most casual conversation



[Web URL for this note](http://notes.benjaminreinhardt.com/Contemporary_research_has_multiple_failure_modes)

[Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/Contemporary_research_has_multiple_failure_modes)




# Corporate RnD orgs have a lot of momentum
Finances matter: corporate budgeting doesn’t redo everything every quarter, instead budgets are benchmarked to a previous quarter or year. So it is harder to get new budget than to keep or expand old budget.
Careers matter: people want to work on visible highly resourced projects. It looks better for your career if your project expands and worse if it contracts or ends. Everybody wants to show a track-record of sustained success all the all the way up the hierarchy. Individual researchers want their projects to succeed so they get out into the world, managers want their researchers projects to succeed so they look good, division level managers want projects that become visible to succeed because that is how they can justify more budget, and the CEO wants projects to succeed to please the board and public.  People want to be associated with success.
There is no natural lifespan of a project. Once a project comes into existence, projects are by default alive.  The budget for multiple projects comes out of a single pool so the number of new projects should roughly decrease with the number of existing projects.
## Related
[[Organizational Momentum]]
[[Sunk-cost fallacy]]
## References
[[Conversation with Charlie Songhurst 12/2018]]



[Web URL for this note](http://notes.benjaminreinhardt.com/Corporate_RnD_orgs_have_a_lot_of_momentum)

[Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/Corporate_RnD_orgs_have_a_lot_of_momentum)




Einstein would have been stuck in a patent office - failures of paradigm-shifting science

You need to break rules to create new paradigms

Bell Labs did not give researchers complete freedom

Comparing and contrasting Bell labs, DARPA, and Venture Research

Contemporary research has multiple failure modes

Is it more impactful to find and train undiscovered diamonds or to unlock well-positioned but constrained researchers?

Paradigm shifts require extraordinary individuals

Eric Weinstein is firmly in the Einstein couldn’t escape the patent office camp

# Einstein would have been stuck in a patent office - failures of paradigm-shifting science
“Einstein would have been stuck in the patent office” is in part driven by how [[Competition and lack of slack  creates specialization]] and how grants and peer review increase the barriers keeping people within discipline buckets. [[The peer review and citation system incentivizes people to work on things that other people think is interesting]]. 

### Related
- [[We have no flying cars - failures of paradigm-shifting engineering]]
- [[Many things cannot be measured well]]
- [[Organizations run by practitioners have less bureaucratic bs]]
- [[To make new things you need to take risks]]
- [[People giving out grants try to derisk them as much as possible]]
- [[The idea that academia is for crazy exploration and companies are for low risk exploration is bullshit]]
- [[Do weird stuff with tenure pledge]]
- [[Making a game more fair reduces its variance]]
- [[Structuring knowledge is expensive]]
- [[Making a game more fair reduces its variance]]
- [[Research has become a commodity]]
- [[Einstein would have been stuck in a patent office - failures of paradigm-shifting science]]
- [[We need more scientific artifacts than papers]]
- [[A system of credit and funding for non-Scientific Paper artifacts should be artifact-type agnostic]]
- [[A system of credit and funding for non-paper artifacts could be a trust network]]

### References
- [[Stagnation and Scientific Incentives]]
- [[odlyzkoDeclineUnfetteredResearch1995]]

[Web URL for this note](http://notes.benjaminreinhardt.com/Einstein_would_have_been_stuck_in_a_patent_office_-_failures_of_paradigm-shifting_science)

[Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/Einstein_would_have_been_stuck_in_a_patent_office_-_failures_of_paradigm-shifting_science)






Current value capture mechanisms are crude

One function of a private ARPA is to fill in the gap once filled by corporate R+D

Corporate labs no longer fill the niche they did in the early-mid twentieth century

We need new institutional structures to fill the role that corporate labs once occupied

The niche corporate labs once occupied still needs to be filled

Could FROs fill the gap left by corporate R+D?

# In the past, corporate labs filled a particular niche in the innovation ecosystem
For obvious reasons the niche is a bit nebulous, but we can perhaps get a sense of it by feeling around the edges and looking at what sorts of things happened at [[Bell Labs]] and [[Xerox PARC]] that don’t seem to be happening now. While corporate labs (Including Bell Labs and PARC themselves) still exist, [[Corporate labs no longer fill the niche they did in the early-mid twentieth century]]. 

[[The niche corporate labs once occupied still needs to be filled]]. 

The trick is to try to distinguish the “what” from the “how.” The What is the immutable characteristics of the activities that we want to try to recreate while The How is the particular set of tactics and strategies Bell Labs used to achieve The What. It’s important to distinguish between the two because we might need to use different tactics to achieve the same things in a new context. If you just blindly followed everything it would be ineffective and [[Cargo cult]]-y.

The other caveat is that there are exceptions to all of these. Startups and academic groups exist that embody these characteristics and not all golden age corporate labs did all these things all the time.

What were the characteristics of this niche? 
1. [[Ideally corporate R+D enables work on a general purpose technology before it’s specialized]]
2. [[Ideally corporate R+D enables targeted dicking around]] - specifically with equipment and collaborators who would not otherwise be accessible ([[Bell labs brought many disciplines together under the same roof and broke down barriers between them]])
3. Corporate labs enabled high-collaboration research work among larger and more diverse groups of people than in academia or startups. 
4. Corporate labs enabled very smooth transitions of technologies along the [[Technology readiness level - TRL]] curve. Corporate labs cared both about novelty and scale. [[Academia incentivizes novelty, not focus]].
5. Corporate labs provided a default customer for a lot of process improvements and default scale for products.
6. Corporate labs often provided a precise set of problems and feedback loops about whether solutions actually solved those problems.
7. Corporate labs provided a first-class alternative to academia where people could still participate in the scientific enterprise. [[Different institutions have created monopolies on various enterprises]]
8. Corporate labs enabled continuous work on a project on a 6+ year timescale. The work to create the transistor took eight years (because of a gap created by WWII) and making it mass manufacturable took another eight. 
9. Corporate labs enabled work in [[stokesPasteurQuadrantBasic1997]] - explorations into natural phenomena with eye towards exploiting those phenomena for human flourishing (and profit.)

I suspect that there is strong overlap between this niche and [[We have no flying cars - failures of paradigm-shifting engineering]]. 

### Related
- [[Could FROs fill the gap left by corporate R+D?]]
-  [[Comparing and contrasting Bell labs, DARPA, and Venture Research]]
- [[A new structure for scalable research]]
- [[We need new institutional structures to fill the role that corporate labs once occupied]]

[Web URL for this note](http://notes.benjaminreinhardt.com/In_the_past,_corporate_labs_filled_a_particular_niche_in_the_innovation_ecosystem)

[Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/In_the_past,_corporate_labs_filled_a_particular_niche_in_the_innovation_ecosystem)






# Internal incentive structures within a large organization work against high quality R&D labs
# Internal incentive structures within a large organization work against high quality R-D labs
Abstractly: [[Organizations reward people for creating new successful products]] so people are incentivized to create something successful. That incentive wouldn’t necessarily preclude working on something with a high chance of failure *but* even if you accomplish something massive, it won’t move the needle a lot at a large organization and you won’t capture a ton of it’s value . ([[The returns to politics is higher at large organizations - Stake v Rank]]). If the only incentive at a corporation is around expected value creation, people will be disincentivized to work on long term high risk projects. 

A  successful corporate (or other) R&D lab requires creating separate incentives from the rest of the org. [[Separate soldiers and scientists]]. My hunch is that modern high-feedback, metrics driven cultures do not do this well. 

The big question is [[What changed from the high-quality corporate labs in the past?]] 

### Related
 * [[Organizations reward people for taking over large projects]]



[Web URL for this note](http://notes.benjaminreinhardt.com/Internal_incentive_structures_within_a_large_organization_work_against_high_quality_R-D_labs)

[Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/Internal_incentive_structures_within_a_large_organization_work_against_high_quality_R-D_labs)




Type I progress is easier to do in developed nations

Most attempts to create Type II progress directly through type I progress fail

Research should be treated like infrastructure

# New technology requires previous infrastructure
Take the “Connecticut Yankee” scenario - if you dropped even the most experienced team of engineers in medieval Europe, they would be hard-pressed to create anything useful without modern tools, and supply chains. 
Additionally, “Process knowledge” - the nitty-gritty subtleties of building things that are near-impossible to transmit via writing may be even more important than physical infrastructure.[^1] 
This infrastructure involves both physical infrastructure - reagents, logistics and labs for biotech, components and manufacturing capability for electronics, and even software requires consistent electric and internet supplies. 
This is why [[Most attempts to create Type II progress directly through type I progress fail]]
[^1]: See Dan Wang’s [How Technology Grows (a Restatement of Definite Optimism)](https://danwang.co/how-technology-grows/)



[Web URL for this note](http://notes.benjaminreinhardt.com/New_technology_requires_previous_infrastructure)

[Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/New_technology_requires_previous_infrastructure)




Activity Space

The linear innovation pipeline is more harmful than useful

In order to turn space into a true frontier we need many technological revolutions

Precommercial scitech is a nebulous term for a certain class of frontier activities

Reasons why an arpa riff would succeed or fail

How do you decide which form a venture should take to maximize its outcome?

New ways of achieving collective action enable new activities

# Overlapping institutional constraints rule out several classes of creative work
Each sort of institution is good at enabling a certain set of activities: academia is good at generating novel ideas; startups are great at pushing high-potential products into new markets; corporate R+D is great at incremental improvements to existing product lines, etc. You could think of each institution ‘covering’ some area on a map of all possible activities. The specific area on the page that an institution covers and the limits of that area are dictated in a complex way by the incentives within that institution. We could call these boundaries constraints on institutional activity. 

Abstractly you could imagine some massively multidimensional [[Activity Space]] that describes the characteristics of different activities that eventually result in awesome stuff.  

You could think of all the institutions together as the ‘innovation ecosystem.’ [^1][^5] Some activities are covered by multiple institutions,  but some aren’t covered by *any* institutions. These activities are ‘constrained out’ of happening by existing institutional structures. For example, unsexy >5 year projects that aren’t focused on a product that don’t produce novel papers  are simultaneously beyond startups/VC (which could support either a shorter timeline *or* more product focus) , Corporate R+D (which would want them to be sexy or product focused), Philanthropy (which would want them to be sexy), and Academia (which would push for novel papers.) Of course there are counterexamples of projects with those characteristics that have been supported by existing institutional structures! It will always be possible to make arguments like “SpaceX exists, so the innovation ecosystem is fine.” Antibiotics are pretty great, even though people can sometimes survive infections without them. Similarly, it’s worthwhile to try to enable more constrained activities even if a few projects make it through. 

Institutionally constrained activity is a more useful and precise way to think about the [[Nebulous]] bundle of things that people label ‘The Great Stagnation’ or ‘research/academia/science/physical innovation is broken’ - the vibe that the world could be on a more wonderful trajectory than it’s on
 right now.[^2]  Thinking in terms of institutional constraints helps us talk about specifically which activities we expect to be happening that aren’t supported which in turn can (hopefully!) lead to addressing those gaps by adjusting incentives in existing institutions or creating new institutions with different sets of incentives. Looking at the innovation ecosystem this way leads me to conclude that [[Contemporary research has multiple failure modes]]: what I would call [[Einstein would have been stuck in a patent office - failures of paradigm-shifting science]] and [[We have no flying cars - failures of paradigm-shifting engineering]].[^3][^4] Addressing each of these areas is important but requires incentivizing different sorts of activities and thus different institutional structures. 

Thinking precisely about which set of disincentivized activities you’re trying to enable is helpful in several ways. It allows you to ask “what incentives are preventing institutions from enabling these activities? Should I be shifting incentives within an existing institution or building a new one? How do I keep a new institution from falling into the same incentive traps?” You can know what *not* to work on despite temptations to do [all the things](https://clickthreetimes.files.wordpress.com/2012/11/do-all-the-things.jpg). ‘Fixing research’ is going to require *many* new institutions (that I suspect will not scale [[Research orgs don’t scale]]) to cover different parts of constraint space. Legible niches can help prevent mimetic infighting over who is going to ‘save science.’ 

### Related
- [[There is a tradeoff between an organizational culture that is good at addressing market uncertainty and one that is good at addressing technological uncertainty]]

[^1]: I hate the term ‘innovation ecosystem’ because it’s usually used as a [[Suitcase Handle Word]], but it is useful to have a shorthand for “all the institutions that enable activities that collectively cover the process of an idea becoming a new impactful idea or technology.” 
[^2]: My unsubstantiated personal opinion is that analyses on high-level metrics like total factor productivity contribute to this unhelpful lumping.
[^3]: I suspect many readers will take umbrage at my fast and loose use of the oh-so-tricky words “research” “science” “engineering” “paradigm-shifting” - here is where I would beg some temporary trust that I take the nebulosity of these words very seriously and have concrete meanings behind my use of them and getting into precise definitions feels like a boondoggle for this particular piece. 
[^4]: This isn’t to suggest that this division is either the only way you can break down the problem space or that these are clean breaks. 
[^5]:And this is completely ignoring non-science and technology focused creative work!

[Web URL for this note](http://notes.benjaminreinhardt.com/Overlapping_institutional_constraints_rule_out_several_classes_of_creative_work)

[Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/Overlapping_institutional_constraints_rule_out_several_classes_of_creative_work)






storrshallWhereMyFlying2021

Metrics can cause paradigm lock-in

Systemic decay theory of stagnation

Lowering barriers to entry is clutch for healthy institutions

Individual rights conflict with societal progress

Once you create a regulation you don't even have to argue anymore

Classical Liberalism defaults towards gridlock

‘Yeah but it’s dangerous’ isn’t a valid argument not to do things

Piecewise intervention doesn't work well for complex systems

Systems engineering was only invented in the 1950s

# Regulations create paradigm lock-in
Regulations effectively lock in a paradigm.  This paradigm lock in can run a spectrum from more explicit to less explicit. The explicit end looks like mandates about specific technology and design choices:“your car must have a catalytic converter.” The less explicit end looks like certain measurements that need to be hit “you can only emit X amount of CO2.” Even measured quantities lock-in a certain paradigm about what is important to measure. From [[Thomas Kuhn]], paradigms `provide model problems and solutions for a community of practitioners, i.e., what is to be observed and scrutinized`.  [[Paradigms outline a set of questions and the rules around answering them]].

A lot of these paradigm lock-ins happen because [[Many things cannot be measured well]] - often we can’t measure safety directly so you need to use proxy measurements that assume a certain paradigm of “what’s safe.” These proxy measurements is why cars are required to have a certain sized crumple zone and thus all have the same shape. It’s much easier to measure physical characteristics of a car than to do a thousand crash tests and see what percentage of them would kill the dummy. 

Regulations also lock-in priority paradigms. The supersonic flight ban locks in a paradigm that prioritizes volume levels near airports over transportation speed. Built into that are implicit assumptions about how loud sonic booms are and a utilitarian calculus about the number (and sort of) people who would be harmed by the noise pollution vs the people who would benefit from having supersonic flight. 

If you take a politics-neutral definition of conservatism as being averse to change, once in place regulations are inherently conservative. (Unless you are conservatively trying to prevent the change that is the regulation itself.) 

[[Regulations slow down technological advancement]] follows naturally through the lens of paradigm lock-in.

### Related
- [[Once you create a regulation you don't even have to argue anymore]]
- [[Changing paradigms is hard]]
- [[Paradigms are lenses for looking at the world]]
- [[storrshallWhereMyFlying2021]]
- [[Labs as regulation-free environment]]
- [[Stagnation]]



[Web URL for this note](http://notes.benjaminreinhardt.com/Regulations_create_paradigm_lock-in)

[Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/Regulations_create_paradigm_lock-in)




Graduate students are the labor in academic science and engineering

Academic credit is a first-to-the post system

People are incentivized to work on nodes higher up the tech tree

Peer review is a great system for preventing waste

Legible novelty is the worst

# The word ‘novel’ is used as an idea bludgeon in academia
“Novelty” is one of the constant criteria by which academic research is judged, both in terms of funding and publication. 

Whether something is novel enough not purely about the magnitude of the contribution (the characterization of reactions that barely differ from other reactions can be considered novel) or the actual newness of the idea (If I argued that the transition between turbulent and laminar flow was mediated by conscious 11-dimensional elves it would be a new idea, but not in the good way.) Instead, the worthy novelty of a contribution revolves around its relationship to existing paradigms.  [[Paradigms outline a set of questions and the rules around answering them]].  

A contribution has massive headwinds if the paradigm judges the question being answered to be already answered or out-of-scope or the answer is not in an approved form (the 11-dimensional elves again.) 

It’s easy to say “argh! The system is full of risk aversion, politics, and narrow mindedness!” But [[Most people are not malicious or stupid]] and that goes for academics as well. Most ideas that answer old questions in new ways or have completely different answers are crackpot ideas (11-dimensional ideas!) so it’s not irrational to default dismiss them given limited time and bandwidth[^1]. 

If a question has existed unanswered for a long time, inevitably many people have tried to answer it. Every failed answer rationally increases the priors that the next answer will also fail. Therefore it’s perfectly reasonable prior that a new answer to an old question will also fail. It’s also technically correct that proposing a new answer to an old question is not “novel” creating a rational excuse for people not to waste their time with crackpot answers. However, the novelty criteria become a cudgel when people use it  to avoid spending the time to actually engaging with an idea. The problem is that it’s not actually clear where to set the filter between “take every crackpot idea seriously” and “reject every answer to long-standing problems.” 

The incentives around new answers to old questions also apply to the creation of new questions. [[Academia pushes against the creation of new questions]]. New questions cannot be measured by their performance on standard metrics or data sets. At least in early 21st century machine learning, it’s easier to get a paper accepted by improving performance on a standard dataset from 98.2% to 9.85% than it is to argue that your system is doing something important independent of an established test or *heaven forbid* has performance that can’t be measured numerically. Again, I suspect this situation at least in part arises from the number of people all clamoring for attention and everyone’s limited bandwidth [[Attention is the ultimate scarce resource]]. New questions can also be completely useless. One out of a thousand times “What if we ran it backwards?” Is the key to saving the universe, but most of the time it’s someone trying to be clever with a bad idea.

These two situations parallel startups heavily. It’s rational to have a prior that a company trying ideas that are littered with the bodies of other companies (memexes, new social networks, elder care, etc.) will fail. However, [[It’s rational to try something that failed in the past as long as you explicitly call out why you will succeed when they failed]] but that requires time and precision from both the founders and appraiser. In parallel with new ideas in science, startups trying to introduce a product that “nobody knew they wanted” usually fail as well.  

The upshot is there are good reasons to dismiss new ideas that are both novel and not novel. Therefore “novelty” is a relatively arbitrary idea discriminator and easy for people to use to dismiss ideas they don’t want to engage with. 

### Relevant
- [[§Academia Constraints]]
- [[Mysteries vs puzzles]]
- [[Peer review has failed as an institution]]
- [[Novelty and Heresy - PG]]

[^1]: The coupling between bandwidth and ability to address crackpot ideas on a case-by-case basis of course plays into my bias that insufficient [[Slack - concept]] is a core source of friction in the system. 



[Web URL for this note](http://notes.benjaminreinhardt.com/The_word_‘novel’_is_used_as_an_idea_bludgeon_in_academia)

[Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/The_word_‘novel’_is_used_as_an_idea_bludgeon_in_academia)




People giving out grants try to derisk them as much as possible

What makes a Florence stop being a Florence?

# To make new things you need to take risks
This is a truism, but why is it actually true? It depends on the assumption that if something doesn’t exist, there is some chance that making it exist will fail and the risk is based on the consequences of that failure, 

### Related
- [[You need to break rules to create new paradigms]]
- [[Knightian Uncertainty]]
- [[Narratives are how we deal with uncertainty]]
- [[Anyone on the knowledge frontier is a bit of a crackpot]]



[Web URL for this note](http://notes.benjaminreinhardt.com/To_make_new_things_you_need_to_take_risks)

[Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/To_make_new_things_you_need_to_take_risks)




# University Patent Thickets - Article
They’re bad
[University Patent Thickets | Research Enterprise](https://researchenterprise.org/2020/09/24/university-patent-thickets/)

### Related
- [[Most tech transfer offices are not profitable]]

[Web URL for this note](http://notes.benjaminreinhardt.com/University_Patent_Thickets_-_Article)

[Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/University_Patent_Thickets_-_Article)






The PARPA model is not original

R+D labs require companies that can win based on Technology

# We have no flying cars - failures of paradigm-shifting engineering
 [[Regulations create paradigm lock-in]] in systems in a way that doesn’t matter to paradigm shifting discoveries. 

### Related
- [[Regulations create paradigm lock-in]]
- [[Regulations slow down technological advancement]]
- [[Regulations have increased by an order of magnitude]]
- [[To make new things you need to take risks]]
- [[Asymmetric career risk impacts corporate R&D]]
- [[University Patent Thickets - Article]]
- [[A new structure for scalable research]]
- [[The vast majority of new ideas and technologies are never reproduced or achieve large-scale impact]]
- [[Academic culture prizes individual recognition]]
- [[The academic incentive structure leads to limitations on accountability, transparency, and collaboration]]
- [[The word ‘novel’ is used as an idea bludgeon in academia]]
- [[New technology requires previous infrastructure]]
- [[Commercialization has very high transaction costs]]
- [[Internal incentive structures within a large organization work against high quality R&D labs]]
- [[Corporate RnD orgs have a lot of momentum]]

[Web URL for this note](http://notes.benjaminreinhardt.com/We_have_no_flying_cars_-_failures_of_paradigm-shifting_engineering)

[Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/We_have_no_flying_cars_-_failures_of_paradigm-shifting_engineering)






Benʼs working notes

We have no flying cars - failures of paradigm-shifting engineering

Related

Links to this note