The Dream Machine

Let me tell you a little bit about my new job

Midjourney, “the dream machine”

I recently started working at Renaissance Philanthropy. It’s a new organization, and most people I’ve met haven’t heard of it.¹ So I thought I’d explain, in my own words and speaking for myself rather than my employers, what we (and I) are trying to do here.

Modern Medicis

The “Renaissance” in Renaissance Philanthropy is a reference to the Italian Renaissance, when wealthy patrons like the Medicis commissioned great artworks and inventions.

The idea is that “modern Medicis” — philanthropists — should be funding the great scientists and innovators of our day to tackle ambitious challenges.

RenPhil’s role is to facilitate that process: when a philanthropist wants to pursue a goal, we help them turn that into a more concrete plan, incubate and/or administer new organizations to implement that plan, and recruit the best people in the world to work on that goal and make sure they get the funding and support they need.

I like to use the Gates Foundation as an example of a really strong philanthropic organization. When Bill Gates decided he wanted to do philanthropy, he did a ton of research, decided what was important to him and what strategies he thought were effective, and built a whole new organization that he leads full-time.

But not every philanthropist is going to go that route. Some donors still work; some want to enjoy their retirement. The default path for donation, the one that takes the least effort for the donor, is to give to an existing, trusted nonprofit organization.

And that’s not necessarily bad, but it does make it hard to do new, bold, effective things.

Most philanthropy is fundamentally steady-state. Whether you’re donating to the opera or to anti-malaria bednets, you’re supporting an existing organization to do pretty much the same thing this year that they did last year.

It’s more difficult — but more interesting — to use your donations to create something in the world that did not exist before.

This can range from things like the Vesuvius Prize for digitally scanning and translating the long-lost Herculaneum Scrolls, to the Cystic Fibrosis Foundation’s “venture philanthropy” investments that funded the development of the first cystic fibrosis drugs back in the 1990’s.

On a smaller scale, ACX Grants are an example of an individual donor funding a variety of early-stage projects (from developing an open-source intranasal COVID vaccine to building a new online prediction market platform.)

It’s intuitive — but surprisingly rare in the overall world of philanthropy — for a donor to go “that’s cool, I want that to exist, it’s a shame there’s no funding for it yet, let me help you get it off the ground.”

And fundamentally I don’t think it’s because donors aren’t generous, or aren’t interested in innovation, but because creating new projects via donation involves a lot of work that mostly isn’t being done.

There’s:

• information work

  • donors don’t necessarily know about cool underfunded projects or talented underfunded people

  • donating to one project you’ve heard of is easy; finding the top 100 projects is a substantial research-and-strategy job.

    • (In philanthropy-land, the job of deciding on a portfolio of people/projects to fund in a given area is called a “program manager”. It’s somewhat analogous to what a VC does in startup-land.)

• negotiation work

  • just because Alice would, in principle, like to donate to Bob, doesn’t necessarily mean the donation will happen.

    • What exactly are the terms of their agreement? What does Alice expect Bob to do in return for the donation? Is Bob willing to do that? They need to come up with a mutually acceptable arrangement.

    • if you’ve ever done complex sales, you know that an informal “sure, I’d like to buy this” is only the beginning of the sales process. If you actually want the check to clear, there’s a lot of people who have to sign off on it, and they’ll all want something a little different, and you’ll have to keep following up and adjusting and persuading to get the deal over the finish line. Philanthropy is no different.

• operational work

  • even something as simple as “Alice meets Bob, is impressed, and writes him a check” requires a little work to make sure the payment goes through, confirm that it’s tax-deductible, etc.

  • if the donor is doing something more complicated — starting a new organization? a prize? a conference, fellowship, grant program? — somebody has to do the literal logistical execution of that.

    • organizations with budgets and employees need hiring, payroll, accounting, etc.

    • prizes and grant programs need to be publicized, to set up an application process, to evaluate applications, and to disburse funds.

    • events need to be planned (publicity! invitations! venue! catering!)

I generally subscribe to the view that most of today’s institutions are too bureaucratic, and that we have too many administrators and too much artificial “process” slowing everything down.

But occasionally people go overboard and demonize all administrative work, or make fun of “email jobs” or project managers, as though you could do without this sort of organizational work altogether. This is unrealistic. Without organizational work — without someone making sure all the pieces come together, which, yes, usually is done through emails and Zoom calls — projects that involve multiple people simply do not get done. I am literally an administrator right now, and I consider it an honorable and necessary role in the context I’m in.

While sclerotic institutions do suffer from too much administration, I’m convinced that a lot of potential ambitious donor-funded projects fail to get off the ground in the first place because of too little (aka zero) administration and infrastructure.

That’s what RenPhil is trying to fix. Philanthropy is work; we do the work.

Science Philanthropy Is Needed

In one way or another, sometimes via my “actual job” and sometimes not, I’ve been involved for a long time with the issue of finding, and funding, extraordinary but underrated science and technology projects.

On the one hand, we live in a world with a lot of bad science — the fraud crisis, the replication crisis, the flood of incremental and unimportant “research” chasing grant funding more than discovery.

On the other hand, even famous scientists are often shockingly underfunded. Despite all the publicity about Irene Pepperberg’s discovery that parrots can learn language, her work is still funded by small donations from parrot fanciers. And, anecdotally, this is not at all uncommon. Just because a scientist has done something extraordinary — even if it’s in newspapers and TED talks and so on — doesn’t mean they aren’t broke.

Right now Dr. Anne Andrews at UCLA has a prototype wearable hormone sensor. I tell people about this, and they’re like “shut up and take my money! I want to know what my hormone levels are in real-time! Where can I buy one?” Well…you can’t, yet. She needs more funding to manufacture enough of these to run a larger clinical study. Somebody’s going to have to donate that.

This sort of thing is all over the place. There are so, so many research projects that don’t exist yet, for want of funding.

Worried about endocrine-disrupting environmental contaminants? Wondering whether the typical doses we’re exposed to cause health problems, or whether it’s just an issue for farm laborers and lab rats who get sick from super-high exposures? Well, the relevant data doesn’t exist yet (correlating concentrations in people’s urine or their foods and household objects with the incidence of disease.) Again, somebody has to fund that!

And, while in principle the government funds science, in practice I have not yet met a single scientist who thinks the process of applying for government grants is fair or practical. It’s not at all unusual for good work to struggle to get traditional NSF or NIH grants. (One particularly insane issue is that most government funding agencies do not pay for professional software engineers, so academics often can’t build production-quality research software.)

One thing many people don’t know is that you can, as an individual, simply donate to a scientist you want to support. Literally all you have to do is email the scientist, get in touch with the university’s grants office, and follow their instructions to send a bank wire.²

Again, the trivial inconvenience of setting this up often blocks people…and if you wanted to donate at scale, to many scientists, the inconvenience would no longer be trivial, but more like at least a part-time job. (And, again, that’s where organizations like RenPhil come in!)

Startups Aren’t Everything

A fair amount of pushback to science and tech philanthropy is the idea that “innovation happens in startups” and “why donate rather than investing?”

I’m highly in favor of startups, investing, and the incentive-aligning effects of the profit motive. But I sometimes meet people who are confused about the actual facts of “where innovation happens.”

Early-stage technology development — R&D — usually does not happen within venture-funded startups, outside of software-only contexts. If we’re talking about biotech, chemicals and materials, hardware engineering, etc? Things that require lab experiments? Then usually the core work is done either in universities, in national labs, or in the R&D departments of large engineering-heavy companies.

Unlike startup founders and academics, who need to “sing for their supper”, publishing journal articles or getting media attention to get funding, researchers at large companies and national labs are a lot more invisible to the average layman. But chemical engineers at, say, 3M, or at the U.S. Army’s research facilities, are just as smart and innovative as their counterparts at top engineering schools. It’s just that their work is often confidential (trade secrets or security classification). But it’s where a lot of “hard-tech” innovation happens.

The constraints of the startup model — aiming for 30x returns in 5-10 years — are a much better fit for commercializing an existing technology than creating a new one. Most hard-tech or biotech startups begin with IP developed in academia. It’s rare to get a VC to fund you to open a lab and run experiments in search of a discovery — from their point of view, that’s too much technical risk, and anyway, basic research is what grants are for!

If you need to spend a few years developing the technology itself, you need a “patient” source of funding, be that a government or philanthropic grant, or a budget within a larger company.

Bell Labs is everyone’s favorite example of innovation in a corporate context, and it’s critical that the parent company did not have to worry about revenue. It was not a startup trying to achieve product-market fit. In the early 20th century, they were the sole American provider of telephone service, something everyone wanted more of; their revenue was, in fact, upper-bounded by the technological ability to deliver phone service rather than by customer demand (which was enormous). That’s why it made economic sense to invest so deeply in their own R&D.

This is the exact opposite of a prototypical software startup, which usually puts together existing technologies into a new product and is testing whether there’s enough customer demand to fuel rapid growth. In a typical startup, R&D is a cost center, not a profit center, and will inevitably be deprioritized until and unless the company has grown large enough to subsidize R&D and wants to make ongoing innovation a priority.³

If you want a particular R&D project to happen, and it hasn’t yet, then you may very well need to fund it with a grant. This doesn’t mean you can’t profit from it — if the inventors do ever get to the point of starting a company, donors will have uniquely early access to become investors!⁴

Here again, inconvenience is a culprit; investors are comfortable with the process of investing, they feel they understand it and they know how to think critically and ensure they’re getting good ROI. They might not feel as familiar or competent in the world of philanthropy, so they’re reluctant to dip their toe in, even in a context where donations make more sense than investments.

Story Time: How Modern Medicis Fail

About ten years ago, I went on a very strange road trip.

I was introducing a rich tech founder to a famous scientist. (They’re both household names, but I haven’t contacted them for permission to tell this story, so I’ll just leave them anonymous.)

The goal was to get Mr. Moneybags to fund Mr. Smartypants to start a new research institute with a radically ambitious goal. Smartypants had laid out his vision publicly; Moneybags had also made public statements that seemed very much aligned with the idea.

I was a graduate student. I was basically nobody. Why was it me in this room?

Well, without getting into the whole web of mutual connections, the idea was that Mr. Smartypants was very hard to get a meeting with, and the point of contact for him was very resistant to the idea to introducing him to anybody who might come across as too pushy and make a bad impression, and I was like “Hi! I’m young, female, and extremely innocuous! I can go!”⁵

So I did go. And the result was disappointing.

Moneybags and Smartypants were strangely tentative in each other’s presence, compared to their fiery rhetoric elsewhere. I struggled to get them talking about their shared goals, or even to create common knowledge that they both thought “science was broken” and needed radical change. Moneybags was deeply suspicious of any kind of not-for-profit organization, while Smartypants was deeply suspicious of business. Any kind of philanthropic/research collaboration between them would have been far outside their usual respective M.O.s, and, indeed, no such collaboration ever happened.

What went wrong?

Well, there are a lot of levels you can look at it. Maybe the idea was no good in the first place; maybe I didn’t facilitate the meeting as well as I could have; maybe something went psychologically wrong with either or both of the two men, making them too timid or cynical to try for the new venture.

But the perspective I find most relevant these days is:

It is INSANE to think that you could broker a multi-million dollar deal to create a new research institute with one meeting led by one graduate student.

Think of it this way. Moneybags and Smartypants already had jobs. Important ones. Making this venture happen was not either of their jobs. It wasn’t even my job; I just spent a free Saturday on it. That’s why it didn’t happen.

For a major new project to happen, it needs to be somebody’s job to make it happen. Several somebodies, really, at this scale.

In the real world, deals like this can take months to get worked out. Somebody (typically you call this a “partnerships” role in philanthropy — the equivalent of sales in business) needs to keep the ball rolling on that negotiation.

New research initiatives need plans. Somebody needs to draw up the mission, the timeline, and the budget for a new research institute before it’s a concrete enough proposal for anyone to seriously consider funding.

New research institutes need staff. Somebody needs to run the operations of this proposed organization. Not Smartypants, he’s a scientist, he might be the “founder” and set the research direction and secure the funding, but he’s not going to be doing payroll and taxes and banks. So who is?⁶

This sort of thing doesn’t just happen ad hoc. It takes work. Back in the day, I didn’t even have much sense of what this work was, and I couldn’t have done it alone even if I had. Even if you believe in running “lean” and avoiding unnecessary staff and process (and I do), to make this vision real would have required any staff and process…which we didn’t have.

Once again, I believe, RenPhil Solves This.

Whose job is it to make worthy projects happen? Ours.

The RenPhil Way

I’m still learning “how we do things around here”, but I’m beginning to get a picture from our founder Tom Kalil (formerly of Schmidt Futures and science advisor to the Clinton and Obama administrations). Fundamentally it makes a lot of sense to me.

1. Anything is possible.

Literally everyone is gettable; no donor, scientist, expert, or authority is “too fancy” to get a meeting with. Plan as though you get to work with anyone in the world you want. RenPhil’s team has spent decades building a network; they actually can get virtually anyone in the world into a meeting.

Nothing is outside the “Overton Window” — there are only projects that haven’t found a home yet. Even if you think something is controversial, expensive, risky, or difficult, don’t rule it out prematurely!

2. Articulate the vision. Concretely. Concisely.

Tom likes to refer to the “magic laptop thought experiment” — imagine writing a press release as though you got everything you wanted. What is the plan you are announcing? Who (remember, anyone in the world, you can name your dream candidate) is doing what role? Then write up that vision.

At RenPhil, the policy is that anything that someone on the team thinks is a good idea, gets written up as a two-page brief proposal. Then, we keep the proposal on file in case a donor comes along who’s interested in exploring it further. Anything worth dreaming of, is worth writing a two-pager on. (More follow-up is appropriate if there’s some indication of an opportunity.)

3. Plans are powerful.

It is suprisingly rare, in this world, to even have a clearly articulated vision or plan. Lots of money and power is sloshing around waiting for someone to come along who actually wants to do something specific with it.

If you’re the kind of person who naturally comes up with research roadmaps, has a long backlog of ideas that “someone should do”, has concrete visions for the future you want to see and can explain why your proposed strategies are a credible path for getting there — Adam Marblestone is a canonical example of someone with this skillset — then you have a valuable talent. There are more people who need an Adam Marblestone than there are Adam Marblestones.

4. Make it easy.

If you want something to get done, then you take responsibility for making it as streamlined and simple as possible for people to help you, even if this means a lot of “handholding.”

One example Tom likes to give is that if he wanted a particular government official to send an email, he would just…send them the text of the email, so all they’d have to do is copy-paste it.

If you need other people’s cooperation in a project, they will flake or delay over trivial inconveniences. Eliminate those inconveniences for them. Spoon-feed them as much as you possibly can.⁷

5. Ask smart people what they think.

RenPhil people ask experts’ opinions all the time. They do it like they breathe.

“Who do you think would be good to lead this project?” “Where do you think this field needs to go next?” “What’s on your wishlist for new projects?”

Asking people’s opinions builds goodwill, it teaches you a lot, and it doesn’t (at all!) mean you can’t form your own opinion independently.

There’s a whole handbook of tactics I’m learning for eliciting opinions so you don’t just get a blank “I dunno.” Things like:

• snowball sampling (when you consult someone for advice, ask them who else you should be talking to)

• starting with examples of the kinds of ideas you want them to generate (i.e. “what other benchmarks and datasets could generate AI-for-Bio advances, the way the PDB and CASP enabled AlphaFold for protein structure prediction?”)

• encouraging them explicitly to dream big and imagine they have infinite funding, can recruit anyone in the world, etc.

6. Be pleasant to all and selective with commitments.

   Everyone is a potential ally, until truly proven otherwise; everyone can be treated with civility and openness to potential collaboration. Then you go home and think for yourself about whether you actually want to work with this person, and only make the (verbal or written) commitments you’re actually willing to follow through on, given your principles and constraints. You don’t have to agree with someone on everything to have a particular goal you want to cooperate on.

   Tom paraphrased a Lord Palmerston quote that I think expresses this well: “No permanent enemies or permanent allies, only permanent interests.”

Got A Dream? I’m Listening.

My main role at RenPhil right now is managing a particular program we’re going to announce shortly — I’ll have another post about that soon — but I’m also keeping a weather eye out for new projects.

I’ve been digging up everything on my own “wishlist” of underfunded projects, from my own decade+ of backlog, and writing up overview proposals and meeting people in the relevant areas.

Looking back over the past decade, some of the things that were “crazy ideas” back then actually have been funded now.

Jean Hebert has an ARPA-H program to replace neural stem cells in the aging brain, for instance. It’s presented in such a matter-of-fact way, but take a moment to appreciate it: holy cow, we are regrowing brains now. That’s a thing we’re doing.

That one guy who could regrow livers? He’s got a clinical-stage biotech company now.

Nothing — at least, nothing that could convince me that it’s plausibly real — is too ambitious or “crazy” to be considered by at least some donors, these days. My sense is that this wasn’t always true, but the way has been paved by our (recent) forebears.

So if you have a science-related wishlist, a person or project that you think needs support, especially if you have a brief writeup explaining why it matters and why it’s possible, I’m all ears.

I’m going to use my own judgment about what I’ll actually try to take action on, but if you’ve been waiting for an opportunity to get some help bringing your plans to fruition, plugging your dreams into a working machine — now is the time to try.

(As always, my personal email is srconstantin@gmail.com).

1. to answer the first question I always get, no, it’s not affiliated with Renaissance Technologies, the hedge fund.

2. by default the university will take about half of your check as “overhead”, but it’s possible to specify that you don’t want them to do this.

3. Think of Google: started with the founders’ PhD thesis; continued to do top-quality CS/AI research to support the development of new products after they had created a money printer with search and ads.

4. and there are lots of ways one could formalize that relationship contractually; I just don’t know much about how it’s done.

5. pro tip: being a young woman who says “yes” to random opportunities can get you into a lot of very interesting rooms — but it only gets you into the rooms. it’s up to you what you do once you’re there.

6. relatedly, if there’s a piece of research that “should” be commercialized as a company, a critical question is who is going to be the CEO? If the discoverer doesn’t want to, or isn’t suited to it, then somebody else has to do it, and effective startup founders are in extremely short supply.

7. (remember, if you’ve ever gotten lots of friendly administrative reminder emails, or clicked your way through a convenient UI, then someone is practicing “make it easy” on you.)