(2021-01-17) Matuschak Reflections On2020 As An Independent Researcher

Andy Matuschak: Reflections on 2020 as an independent researcher. 2020 was my second year as an “independent researcher.”

1. Why work independently?

I’m interested in inventing environments which significantly expand what people can think and do. (enabling environment)

If the work goes well, an independent researcher will likely find compelling opportunities to evolve into some higher-leverage institution—a studio, a foundation, an academic center, a business

I want to more deeply understand the properties of enabling environments—principles of operation, design procedures and patterns, relationship to individual and social cognition—to foster a community which routinely invents new environments of this kind.

Why not start a startup?

my interests are misaligned with a startup’s fundamental drive: growth. (Neo-Victorian)

Some companies figure out how to align these aspirations so that marginal revenue/usage enables marginal fundamental insight, which enables marginal revenue/usage, and so on, in a virtuous cycle. Pixar is a good example. Cutting-edge graphics research enables new kinds of storytelling, which in turn funds more research.

Such flywheels are particularly rare in my domain, since novel interface ideas are typically public goods. Their development might require significant investment, but once created they can often be copied cheaply by others, limiting returns to the inventor.

A related question I’ve had to straighten out for myself: am I really aiming for insight, or for impact?

I wouldn’t want to work on a project without the possibility of transformative impact. But so long as I have some persuasive theory of how that impact could happen, I prefer to focus on producing insight through prototypes.

a useful way to frame my goal: developing ideas far enough that they become “obvious,” the banal fodder for half a dozen companies in a future YC batch.

Another key misalignment makes me hesitant to consider a startup: culture. Tech culture is different from research culture, and I’m already quite overweight on tech culture

For example, tech culture is calibrated to a much faster pace than research culture

It’s not that the tech people are constitutionally lazy or something like that: in industry, it usually is, in fact, a bad idea to spend many hundreds of hours thinking about a single problem. Better to create an 80/20 solution or try a different approach. But foundational insights often do require more patient, focused thought

Why not become an academic?

the real reason is my choice of field. My goals, values, and practices align poorly with the academic discipline which would most naturally host my work, human-computer interaction (HCI). If, in some alternate universe I were interested in a different topic, there’s a good chance I’d have become an academic.

A subjective and inflammatory way to put it is that I feel academic HCI discourages vision and imagination. Part of the trouble seems to be that the field’s trying too hard to be a science.

I read the major conferences’ proceedings every year, yet I learn much more about HCI from studying game designers and idiosyncratic Twitter tinkerers.

It feels like skimming a sea of churning froth—tiny isolated studies rarely accreting into a broader current.

So I work independently. Not because that’s an ideal arrangement, but because I don’t see a good alternative. I don’t yet know how to create or join an institution which would enable better work

I don’t understand yet my own projects well enough to effectively coordinate a large team around them

2. The unexpected success of crowdfunding research

My collaborator Michael Nielsen suggested we set up a Patreon to solicit funding for our work, and I agreed, not thinking much of it. He’s now moved on, but I’m quite grateful to him for that early nudge.

Less than two years later, my patrons have crowdfunded roughly a graduate student’s fellowship grant. It’s not lucrative, but it’s enough to cover my living expenses.

Growing a crowdfunded research grant

First, we should examine the dynamics of growth and churn.

On my Patreon, I offer three funding tiers: $5, $20, and $100. The tiers are mostly arbitrary for now—the only difference is that the $100 tier comes with public attribution (which no sponsor seems to care much about). But the tiers give people a way to modulate their support according to their interest and capacity.

At Khan Academy, we were often quite dependent on a small number of huge-value donors, and we’d regularly fret about appeasing them

Researching in public

The promise of ongoing exclusive content creates some tension, then. Unlike a typical paid newsletter or blog, funder-exclusive writing is a secondary by-product of my primary work

As an antidote, I’ve tried to engage in what Michael has called “anti-marketing”. That is, to make a point of focusing publicly on the least rosy parts of my projects—what’s confusing, what’s frustrating, what’s not working. It’s hard to do consistently, but when anti-marketing is the goal, then interesting challenges become something positive: useful fodder for public conversation.

More broadly, I’ve experimented this year with a mindset I’ve been calling “working with the garage door up.” (Work With The Garage Door Up)

This has worked quite well when I adopt the right mindset—that I’m sharing objects made as part of my primary work, rather than things created specifically for publication.

The practice generates more conversation and serendipitous inbounds “for free.” It’s worth noting that in most ways, unusual inbounds are a better leading indicator for my work

But if you’re doing something original, the most interesting aspects are the ones which others—and you!—understand least well

Others’ replies will tend to emphasize the most mainstream elements, since they may not notice or know how to react to the aspects you least understand. Such conversation will often drag you back towards the mainstream. It’s a kind of “regression to the mean” for ideas.

Great colleagues and collaborators can take more active steps to mitigate the issue.

What are patrons buying?

In my interactions with patrons, I’ve been surprised to find that altruism is rarely the dominant force. Patrons mostly don’t think of themselves as paying for consumption of past work; they’re buying into production of future work.

future work to be produced which would not have been produced without it.

About a third framed their funding in terms of “people, not projects,” expressing general confidence that I’ll do interesting work.

My path to sustainability

How repeatable is my modest success with crowdfunding?

First and foremost, I was in a position financially to draw no income for (as originally planned) several years

My wife’s generous understanding was unusual and essential. Practically speaking, though, it helps that she’s a doctor.

Before I left Apple, my plan was to save enough that I could quit and pay myself a grad student’s stipend from the interest indefinitely. It would have taken another five years or so, depending on my diligence. That seems very achievable, though I was so uninspired by the work at Apple at that point that it might have done some permanent damage!

I avoided burning through savings by becoming the lucky recipient of an Emergent Ventures grant. It covered my first year’s expenses. I would certainly recommend Emergent Ventures to others trying to find a way to support their independent work. The application took only a couple hours; I enjoyed a half hour of thoughtful conversation with Tyler Cowen; they made a decision within a few days; my only obligation was to send them a few short reports.

Finances aside, a few related career capital factors have likely been crucial to my progress. They may be important for others doing similar work.

First and foremost, I’ve had enough professional experience to build up a rare skillset. I can independently research, design, implement, analyze, and report on novel software environments.

My past experiences have also yielded essential social capital. Most people don’t seem to care that I’m unaffiliated, but I suspect that’s only because I can introduce myself by saying I helped build iOS at Apple and led R&D at Khan Academy. Without some kind of strong social proof, it may have been hard to get anyone to engage seriously with my work.

I’m much less sure about this last point, but I suspect my work depends on time spent living in San Francisco

Patrick Collison advises: “Figure out a way to travel to San Francisco and to meet other people who’ve moved there to pursue their dreams. Why San Francisco? San Francisco is the Schelling point for high-openness, smart, energetic, optimistic people. Global Weird HQ.”

Living here has changed me deeply.

Can you absorb a scene’s values through the internet? I certainly did to some degree as a teenager living in Saint Louis.

But my sense is that bandwidth limitations in interaction are significant. There are depths which are hard to absorb without constant physical immersion

3. Working alone

I may be working independently, but that doesn’t have to mean working alone

Executing alone

how much capital do I feel I could productively deploy towards my goals (annually, say)? In early 2019, it was a few tens of thousands; now I’d put the number at a million or two.

I now understand the work well enough to imagine how I could accelerate it with a team

there are prototype ideas which I don’t explore because it would take too long

My approach requires developing new software interfaces which express insights, then studying those interfaces and their use to generate new insights, and so on

In practice, it’s quite difficult to think deeply about theories while in the midst of a significant software development project. They’re different states of mind

I’ve not had much success when dividing my days or even my weeks into “building” and “thinking” blocks.

Maybe that’s fine. I just need to spend months at a time in one mode or the other, and I need to get more patient.

I’m worried that engineering mindset stunts my growth as a researcher (still quite nascent), even in the weeks following an intense period of implementation

Another problem with cycling slowly back and forth is that feedback loops become too long. (bootstrapping)

In 2021, I’d like to experiment with volunteer and (funding dependent) contract collaborators

Culture and scenius

I may not want to join academia, but I deeply envy a good field’s intellectual kinship

I’m grateful for intermittent conversations with others doing related work, but for the most part it’s not enough

I want to be part of a rich scenius of serious, capable people doing full-time original research on enabling environments.

I want peers who will candidly observe the limitations of my ideas, then work with me to improve them

My proto-field has (hopefully) a proto-scenius: many part-time tinkerers, many startups doing research-ish work when they can spare the time

Serious contexts of use

The most serious problem for me in working alone is that it means I lack a serious context of use.

Practically speaking, such contexts provide deeply meaningful feedback.

But perhaps most importantly, these projects also provide the intense personal connection which makes great work possible.

one recipe for insight through making might be deep collaboration between some colleagues focused on some serious domain problem which might benefit from augmentation (“tool-users”), and other colleagues looking to use that context to drive the creation of new environments (“tool-makers”). My plan for 2020 was to start an experimental media studio along these lines with Michael, but I shelved that approach when his plans changed, since it depended on his skills as a synthesist

Is my position a fluke or an example? Can independent researchers be coordinated into a scene without traditional institutions?