sawyer

Sawyer is the Executive Director of the Berkeley Existential Risk Initiative (BERI). Prior to joining BERI in July 2019, Sawyer was a Production Manager at Research Electro Optics (now Excelitas Boulder) and Head of Production at Modular Robotics. He has a B.S. in physics from Tufts University.

Wiki Contributions

Comments

You Should Write a Forum Bio

I just fleshed out my bio a bit more, thanks!

AI Safety Career Bottlenecks Survey Responses Responses

Thanks for publishing this! Are you planning to release a list of wishes/bottlenecks to which you didn't have a response? 

Announcing "Naming What We Can"!

It's BERI fun to make puns like this, they've always been BERI popular among BERI employees.

Announcing "Naming What We Can"!

In accordance with best practices, BERI will be changing our name from "Initiative" to "Institute" and our acronym from "BERI" to "BERI 1".

AMA: We Work in Operations at EA-aligned organizations. Ask Us Anything.

There is probably a version of this that works. I know that some orgs have tried something sort of like this. One problem with "hire contractors who aren't value-aligned to do various things" is that you then have to manage those people (whether they're value-aligned or not). This is getting outside the scope of your original question, but I think people underestimate how much it can drag on an organization to hire new people without thinking hard about their onboarding. If you have a very well-defined task, with a clear deliverable, then it can make sense to contract it out. But if you just have a problem and need help, often bringing on a new person to solve that will make things worse in the long run.

AMA: We Work in Operations at EA-aligned organizations. Ask Us Anything.

One general answer and one specific answer. (Numbering is my own and doesn't correspond to your questions.)

  1. When implementing a new system that will be used by a variety of people in your org, it's important to make this as easy as possible for them. As Martin said, if you end up with one person who doesn't want to use the system, this will greatly increase your work.  When you're choosing a new tool or system for people, you should assume that all of your new users don't care about it, and that any time they have to spend learning the new system will be time they'd rather spend doing something else. To that end:
    1. Don't just ask people to read the software's own published user guide. Even if it's really great and you couldn't imagine making anything better! People don't want to click on links, they don't want to go somewhere else and see new branding, they just want you to tell them what to do.
    2. Do write specific instructions, in an email, as clearly and concisely as possible. Even if those instructions mirror the software's user guide, it's likely you can make them more concise because your instructions are for your org's specific use case. 
    3. Do offer to help anyone who struggles with the new system. It's unlikely anyone will take you up on this offer, but I think it makes people feel like you know they don't want to do this thing you're asking them to do.
    4. Be ready for people to keep trying to use the old system. Keep trying to shepherd them into the new system. Don't give up, don't resign yourself to the two-system world. People will fail to see your first email, but happily respond to your follow-up. Seize opportunities to walk someone through a concrete example: Maybe they didn't set up their Expensify account the first time, but now they're asking for reimbursement, so you can (nicely, graciously) force them to use Expensify to get it.
  2. BERI uses QuickBooks Online (QBO) as our accounting software. It works well enough, but has a lot of quirks that make certain things much more difficult and confusing than they have to be. Consequently, I'm (slowly, hesitantly) looking into other options, Aplos in particular. Obviously I'll want to try out all of our normal transactions and reports in Aplos before transitioning. But even if all of that works out really well, one thing I'm very worried about is loss of audit trail: QBO keeps a record of every change to every transaction, including the time, date, and user. I've only used this occasionally, but when I have it's been a lifesaver. I'm not sure how I'll get around this if we start using Aplos, but it's definitely something I'll be keeping in mind. This concern would apply to any important piece of software that stores a complex history of interactions. 
Effective Altruism Coaching 2020 Annual Review

I see, thanks! Any ideas why? Maybe just because you have more clients now, and larger groups tend to have lower response rates for this sort of thing?

Effective Altruism Coaching 2020 Annual Review

Great write-up. Thanks for doing this, and thanks for your work!

You mentioned that "Over the past year and a half, a smaller fraction of people responded to the survey," but I don't think you report what those percentages are? The closest I can see is "200+ clients" and all the various n values, which seem to peak around 60. But you don't report your current number of clients, so I can't take much from these data. Can you report approximately what fraction of people responded in 2020 vs previous years?

Money Can't (Easily) Buy Talent

I don't consider Open Phil to be an example of Earning to Give. My understanding is that basically all of their funding comes from Dustin Moskowitz's Facebook stock. He completed his work on Facebook before taking the Giving Pledge, so his primary earning activities were not chosen in the spirit of Earning to Give.

It's also not clear to me that the EA Funds are examples of EtG. The EA Funds take frequent donations, and my impression is that they have many donors. At least, I don't see any evidence that the donors are purposefully Earning to Give (i.e. that they chose their jobs as a way to maximize earnings with a plan to donate).

It's possible that you and I have different definitions of EtG. Mark's post doesn't explicitly define it. Wikipedia's definition does not seem to include "normal" donors who give, say, 10% of their not-super-large income.

These examples might not be critical to your first point, but I think you would need to provide other examples of grantmakers that are more obviously funded by EtG (e.g. by evaluating Matt Wage's personal grantmaking).

Money Can't (Easily) Buy Talent

I think that overall this is a great post, and that you've made serious progress towards concretizing some vague concerns I have about EtG.

For me, the most striking point was:

Given the current lack of infrastructure to make use of unaligned individuals, the benefits of hiring such people is low. This analysis recommends building infrastructure that allows for money to be used more effectively, e.g. increasing the management and training capacity of existing organizations.

I had not heard this before reading your post, and it feels novel and useful to me. I don't think it's true for all roles, but I like it as a way to think about some roles.

Two things in the post confused me.

First thing:

However, due to communication and managerial overhead, two people will do less than twice the work of one person or cost more than twice as much. One person with twice the talent is more valuable than two people, just like an iPhone 15 is more valuable than two iPhone 12s.

The purpose of hiring two people isn't just to do twice the amount of work. Two people can complement each other, creating a team which is better than the sum of their parts. Even two people with the same job title are never doing exactly the same work, and this matters in determining how much value they're adding to the firm. I think this works against the point you're making in this passage. Do you account for this somewhere else in your post, and/or do you think it affects your overall point?

Second thing:

You use the word "talent" a lot, and it's not clear to me what you mean by that word. Parts of the post seems to assume that talent is an identifiable quantity, in principle measurable on a single scale. I think that many (most?) real world cases don't work like this. To me, "talent" is a vast array of incommensurable qualities. Some are quantifiable, some are not. In practice,  the market attempts to rectify this by (implicitly) assigning monetary value to all these quantities and adding them up—your post argues convincingly that it regularly fails to do so.

But if we can't really even measure talent to begin with, what are we even talking about when we talk about talent? What do you mean when you say "talent"?

Load More