# What spins the innovation flywheel?

<div data-node-type="callout">
<div data-node-type="callout-emoji">💡</div>
<div data-node-type="callout-text">Please mind: this is an <strong>extremely reductionist </strong>PoV, and also open to discussion and criticism. Also, it’s an <strong>undereducated</strong> PoV - I’m learning on the go, and will propose a <strong>modelling/simulation appproach</strong> shortly.</div>
</div>

# What drives progress?

I’ve always been plagued with the “how do I create more good?” idea, for the good or for the bad. And I’ve been thinking about it, too <s>(some say thinking too much is the bane of one’s existence…)</s>, stumbling upon [80000 Hours](https://80000hours.org/) some time ago. Their ideas, while awesome, didn’t quite click (say, 80% click vs. 100%) - so I continued to think some more…

<div data-node-type="callout">
<div data-node-type="callout-emoji">💡</div>
<div data-node-type="callout-text">N-th order effects → cascading, long-term consequences of incentives. Just in case.</div>
</div>

One question I’ve been asking myself - **what is driving the progress?**

## Oof

Obviously, it’s **curiosity** and the desire of knowledge, yeah - but now, are those the only factors?

![](https://cdn.hashnode.com/res/hashnode/image/upload/v1737828634426/948aa013-c3e0-4df9-83d5-bde342970a3e.png align="center")

I’d also say current **political situation, business and science** also have a go.

![](https://cdn.hashnode.com/res/hashnode/image/upload/v1737965160375/84a496be-33a8-46fa-b1cb-f81206a2084b.png align="center")

## Deeper oof

![Lobbying-Good-Bad-03](https://represent.us/wp-content/uploads/2016/03/Lobbying-Good-Bad-03-1.jpg align="center")

Let’s dive somewhat deeper (DELVE DELVE DELVE I’M NOT AN AI) - **politics and science are influencing progress and innovation**, yeah.

![](https://represent.us/wp-content/uploads/2016/03/LobbyingROI_Update-290x300.jpg align="center")

But aren’t [**politics driven partly by big corporations**](https://represent.us/action/is-lobbying-good-or-bad/) **and other stakeholders, and science - by the same ones, in the form of grants, endowments** etc? Let’s add **capital** to the chart:

![](https://cdn.hashnode.com/res/hashnode/image/upload/v1737965379821/60ee502e-407c-4794-b831-00258e4e59f9.png align="center")

## So deep I can’t breathe

![The academic research funding system: as it is now | by Coalfacer | Medium](https://miro.medium.com/v2/da:true/resize:fit:1200/0*_iSAL9ucU0aflp38 align="left")

Yet, to receive lobbying contribution, raise a VC round or receive a grant you’ve got to:

* be in **high-profile politics**
    
* be a [**serial entrepreneur**, or a **top university graduate/PhD**](https://freedium.cfd/https%3A%2F%2Fblog.creandum.com%2Fdo-vcs-add-value-4-years-later-the-answer-remains-sometimes-49c53a8865b4%3Fgi%3D47fe533bb123)
    
    ![Do VCs add value? 4 years later, the answer remains: sometimes. | by Carl  Fritjofsson | Creandum](https://miro.medium.com/v2/resize:fit:1400/1*_Gt7Lp6n5CK4Jevlicn85g.png align="left")
    
* **enter a good/great laboratory** with a PI that [gets the **grants**](https://www.timeshighereducation.com/news/winners-and-losers-in-hefce-funding-allocations/2019306.article) <s>(imagine a growling PI that jumps on top of others to snatch the grant approval paper)</s>
    
    ![List of Top Research Universities in The World By Funding | World Economic  Forum](https://assets.weforum.org/editor/GdRN2GWqcJqMCXt482fiWO6oOzAtv3oE9DnHUO0Ys0U.jpg align="left")
    

(Had to change D2’s layout engine to not make a mess…)

![](https://cdn.hashnode.com/res/hashnode/image/upload/v1737965512564/72f6aa3c-58af-4bf0-99f6-787dde472813.png align="center")

# How can we foster and promote innovation?

So, quality education increases the expected value of science, business, capital etc.:

How can we make education:

* better
    
* more targeted
    

at driving humanity forward?

IMO, some possible intervention directions come to mind:

* selection based on natural traits, i.e. [talent identification](https://www.perplexity.ai/search/which-countries-choose-childre-Gg9rbsflRgWr3k.nuWSjig) → low-nepotism progress in universities and PhD programs
    
* [**project-based learning**](https://www.perplexity.ai/search/project-based-learning-effects-.qWBrA3KSL.JTZRU.hBIuw) starting with school, continuing further
    
* science that’s more open (less [publisher domination](https://wordsrated.com/academic-publishers-statistics/)) and more decentralized (P2P reviews?)
    
    ![Academic Publishers Statistics – WordsRated](https://wordsrated.com/wp-content/uploads/2023/06/Number-of-academic-articles-published-top-10-1024x683.png align="left")
    
* showing potential problem areas to direct curiosity (without aiming at particular problems, yet researching them - we don’t want to instill [einstellung](https://en.wikipedia.org/wiki/Einstellung_effect) effect into anyone)
    

## Incentivize me, daddy

For that to happen, policies have to change. For policies to change, right incentives have to be imbibed into the system: you’d want to incentivize the ones designing and approving policies first, for the **n-th order effects to unfold through the layers** like mighty ducklings making waves (remember, this is HUGE):

![Kyle Roth on LinkedIn: Physics textbooks: "Solve for wavelength of the duck"](https://media.licdn.com/dms/image/v2/D5622AQF6fjZ1dmLTRQ/feedshare-shrink_2048_1536/feedshare-shrink_2048_1536/0/1687357732237?e=2147483647&v=beta&t=Na2YWdv_F_imLveODWg7XGDzbgt4NQFo1d4vespFXFM align="center")

Which leaves us with something akin to a flywheel where we **fine-tune incentives for policymakers → educators → businessmen and scientists:**

![](https://cdn.hashnode.com/res/hashnode/image/upload/v1737832431918/e5e404c1-11d2-48d5-9bde-1a9564f9af38.png align="center")

Fine-tuning can have [different goals](https://www.inffer.com.au/resources/public-private-benefits/):

* **optimal policy adoption** → e.g. changing policies will hinder both their efficiency and the time to collect data for benefit evaluation
    
* **net benefit optimization** → the end goal. <s>Someone should definitely read more on </s> [<s>public policy analysis</s>](https://fastercapital.com/content/Public-Policy-Analysis--The-Role-of-CBA-in-Shaping-Effective-Policies.html)<s>…</s>
    
    ![73 – Public benefits, private benefits, and the choice of policy tool –  Pannell Discussions](https://www.pannelldiscussions.net/wp-content/uploads/2011/12/pd73f3.gif align="center")
    

## How can all that be measured?

So, we need effective policies to create this flywheel, but *how do we know if they are working? How do we measure their impact and optimize them?*

<s>I have no idea, and I feel like a low-parameter LLM regurgitating smart words.</s>

Maybe (a high-level uneducated assumption) a bandit-like framework can help, where:

* several hypotheses are tested
    
* benefit effect and [intervention efficiency is assessed](https://resource-allocation.biomedcentral.com/articles/10.1186/1478-7547-10-8) (e.g. [causal inference](https://tnsr.org/2020/07/applying-method-to-madness-a-users-guide-to-causal-inference-in-policy-analysis/) with counterfactuals)
    
* [optimizing and optimizing](https://freedium.cfd/https%3A%2F%2Fmedium.com%2F%40fellipe_marcellino%2Fmulti-armed-bandits-comparing-different-algorithms-c5925eed87b) again
    

P.S. Yeah, this n-th order effect thinking was partly the [reason for my blog naming](https://en.wikipedia.org/wiki/Teleogenesis): we, as virtual higher-order policymakers (incentive is a [particular case/factor of a policy](https://en.wikipedia.org/wiki/Incentive_program)), want to create teleogenic frameworks so that incentive optimization process would be self-starting.

P.P.S. And yeah, we should definitely check for [Goodhart’s law](https://en.wikipedia.org/wiki/Goodhart%27s_law) creeping in - complex system optimization, akin to [overfitting](https://www.ibm.com/think/topics/overfitting), paves a way to unintended consequences:

![Too much efficiency makes everything worse: overfitting and the strong  version of Goodhart's law | Jascha's blog](https://sohl-dickstein.github.io/assets/cartoon-mid.png align="left")

---

Welcome to **Teleogenic**❣️

Other places I cross-post (not always) to:

* [**Hashnode**](https://posts.teleogenic.com)
    
* [**Medium**](https://baldr.medium.com/)
    
* [**Telegram**](https://t.me/ohmyboi)
    
* [**Twitter**](https://twitter.com/ZakharKogan)
    
* [**LinkedIn**](https://www.linkedin.com/in/zakhar-kogan/)