Freedom at PSE

# On research freedom at PSE... How much is too much? How little is too litte? > Too much of everything is just enough > -Jerry Garcia As much as my dad loves the Grateful Dead, I'm skeptical we should listen to them... ### 🌶️ Sam's spicy take: PSE researchers currently have too much freedom on what they work on ...Ok, interesting "hot take", Sam... ...Not very surprising coming from "management"... ...but why is this topic even worth talking about? ...Doesn't PSE have a bunch of more important & urgent things to figure out? ## Why is research freedom an important topic to debate? Some folks have argued that “Freedom in research topics” is an operational issue & that this topic is not a high priority: > [these operational issues are not the most urgent things to discuss right now](https://hackmd.io/wvD5httWRnuM62kl1fMduA?view) Perhaps "How much freedom should we have in research topics?" is not THE most urgent thing to discuss, but I want to make the case that: **this question is not merely an operational issue, it’s a key strategic issue to debate, because it gets at the core question of "how should we approach research here at PSE?"**. It directly relates to other questions such as: - how much collaboration should we have within the research team? - how much cohesion should there be across topics for the research team? - how aligned should the direction of the research team be with PSE as a whole? - to what degree should research priorities be driven by development needs? - to what degree should development roadmaps be driven by new research output? I believe these are all key strategic questions. ## Hard truths We must acknowledge the inherent tradeoff between individual autonomy & internal collaboration. I’m sensing a key tension here within PSE that we should surface, deliberate & make a decision on. Looking at outputs from the recent research workshops, it’s clear that freedom of research pursuits is highly valued within the team > [Participants emphasized PSE’s strengths: the freedom researchers have to pursue meaningful topics](https://www.notion.so/pse-team/Summary-of-workshop-Group-A-114d57e8dd7e80b59513ca006fb179c6?pvs=4#114d57e8dd7e802fb2eff8b44a755b05) > [People mostly value the freedom and autonomy that researchers are given to pursue their own interests without having any strings attached](https://hackmd.io/@letargicus/H1HZxk300#State-of-the-Art) Broadly, I support us all having a degree of autonomy. Clearly some amount of freedom is critical. People should enjoy what they’re working on. They should be passionate & excited about their work. It should bring meaning & purpose. But I wonder, **how much freedom** should researchers have on what they work on? Complete, unbridled freedom? ...I don’t think anyone would argue for this (but feel free to comment here if you do). So the question becomes: on the spectrum of complete individual freedom vs. complete top-down control, where should we place ourselves? **What are the right of amount of constraints to enforce?** I’d argue: - Not too many constraints to remove joy & passion from our work - Not too many constraints to stifle creativity (though I believe constraints can help encourage creativity) BUT it's worth recognizing there's serious downsides to enabling too much freedom For instance, read [this post from Enrico](https://hackmd.io/@letargicus/ryR85a-6C), who argues we lack internal engagement: > Internal engagement means the opportunity to discuss your research problems with people who are well-prepared and opinionated on that particular topic, who can provide help and/or challenge your assumptions. Ideally, such discussions should bring doubts and uncertainties to the table but also ideas and avenues to explore. They should be intellectually painful. Instead, most of the time current discussions are passive and lack of engagement. I think this is precisely a result of our team(s) having too much freedom to work on whatever they choose. As a result, no one else has a deep understanding about what you're working on. No one can challenge your ideas, or provide feedback on how you can improve. Perhaps that gives you freedom & prestige, but now we're stuck with passive discussions that lack engagement - that doesn't sound like a very fun or exciting place to work. [Enrico digs into this further](https://hackmd.io/@letargicus/H1HZxk300): > Giving researchers total freedom naturally leads them to choose the most disparate and unrelated to each other problems Without cohesion on what we’re working on & what we're aiming towards (i.e. a unified PSE mission & vision), this results in people working on very different things. If you work on something very different than the rest of PSE research, you become the sole expert on this topic. No one else has the ability to evaluate your work, give you guidance or input, challenge your ideas in order to make it stronger & to move it in a better direction. Similarly, PSE development projects likely are less interested in what you're working on, because it's less likely to be able to solve their problems. Your work is likely less relevant to the rest of PSE. This leads to a lack of collaboration... > [people identify the lack of internal collaboration within the team as the most pressing problem](https://hackmd.io/@letargicus/H1HZxk300) > ...as well as a lack of engagement. No one else has any clear motivation or incentive to care about what your working on. Is that the world we want? Do you really want to work hard on something for 6 months, just to produce a result that no one cares about? **Wouldn’t you be more excited to see PSE project teams HIGHLY interested in what you’re working on? People who are asking questions, eagerly awaiting the output of your work, in order to incorporate it into their codebase, and to bring concrete impact to real people out in the world??** I would hope our researchers are ultimately more excited about this potential for concrete impact, rather then being here merely for the intellectual feat of solving a hard problem. As CC put it, it’s one thing to be an inventor. It’s harder to be an innovator: > [The distinction of "Invention" and "Innovation" is helpful for the discussion. Everything produced from the R&D departments are inventions. But only turning the invention into useful social product we can then call it "innovation."](https://hackmd.io/0YgI9Y4mTamjge7bT-KHkQ#Pull-and-Push) Or as Laksham put it, are you here to build technology? Or are you here to build art? > [it's not technology to me unless it’s useful](https://ls.mirror.xyz/4znIcGEQG3mVDJjZ_7Eg5uWxon-eJqJCgoHIwILWJhg) ## Push vs. pull I strongly agree with [CC's take on push vs. pull](https://hackmd.io/0YgI9Y4mTamjge7bT-KHkQ#Pull-and-Push) approach to R&D. It's a mindset shift. Rather than approaching this massive technology landscape asking "here's all this tech, what can we do with it?", we should be orienting around users we care about & their concrete problems, then asking "what tools do we have available to solve these problems?". ## Building a sum greater than our parts With the current amount of freedom we have available, there’s no advantage to us being a "PSE research team". We might as well be a bunch of independent academic groups that share the same (virtual) office. Maybe there's some occasional water cooler discussion & areas of overlap, but we're mostly isolated. We’re merely the sum of our parts. > [Some could argue that we are not even a proper team but just a sum of clever individuals doing their own stuff.](https://hackmd.io/@letargicus/H1HZxk300) Instead, We SHOULD work to create an environment (via cohesive technical direction & culture) where the whole is GREATER than the sum of our parts. So what should we do? Again, I strongly agree with Enrico: > produce a compelling tech thesis everyone believes in. Under such circumstances, the collaboration graph would look much denser with tight and powerful connections between researchers Easier said than done, for sure, but I agree worth striving for. I will share more ideas on PSE mission & vision at Precon, which I think will help us focus our tech thesis more narrowly. I also look forward to the Precon research discussions on this! ## How aligned should R&D initiatives be? Should research findings drive the roadmaps of development projects? (push) - e.g. should all/most new development projects evolve from new primitives that emerge from research efforts? Should the needs of our development projects drive research priorities? (pull) - e.g. should all/most research efforts be focused at solving specific existing problems for our development projects? I think it’s at least worth the thought exercise of taking the extremes & discussing tradeoffs of the various approaches here. It's ultimately a balancing act, but my gut tells me that **more of our research time, energy, & resources should be dedicated to challenges our development project teams face, which should be oriented around problems our users actually care about**. Here’s a hypothetical — what if we established these as the top organizational priorities for PSE? - Priority A) - **Within 1 year, run a MACI poll in production (e.g. on Ethereum L1/L2) that has completely removed the trusted coordinator** - How would we organize our research efforts to meet this need? Would we explore certain approaches of MPC and/or FHE and/or TEE more earnestly? - Priority B) - **Within X timeframe, remove all notary trust assumptions in TLSNotary** - Again, how we we (re)organize to meet these needs? - Priority C) - **Within Y timeframe, reduce client-side proving time for Anon Aadhaar by 90%** - Again, how we we (re)organize to meet these needs? What if we organized all our resources & activities around these goals? - How would this impact the research efforts? - How would researchers feel about these as priorities - Would you be excited by the challenge? - Or revolt at the lack of freedom? Comments welcome! I’m genuinely keen to get input on this. How much should research prioritize the needs of development projects? - Should it be 100%? No - Should it be 0%? No - Should it be more than what it is now? I think yes What is the right %? Is there an ideal balance? I think the right answer is likely a mix of both: - Most of PSE research should aim to solve real needs for our development projects: - Removing trusted roles from protocols - Reducing client-side proving times for applications - Improving developer productivity/performance/security to build & deploy cryptography applications - Some of PSE research should have the freedom to dream big, to explore areas of uncertainty, that may not yield tangible results for years to come, but have the potential to massively unlock new capabilities How does a researcher earn the right of freedom? - Is it a given? - Is it tenure? - Is it proven success in previous PSE research efforts? - Is it recognition of one's gift of "taste" or brilliance, identified by "leadership"? I look forward to exploring these questions & more at Precon :)