So you've decided to hire Caroline Ellison. How should you announce it?

Two days after announcing that Caroline Ellison had joined Manifund, Austin Chen asked CEA's community health team to 'please let me know if there are things you believe I should do moving forward to mitigate this impact.'

I attempt to answer this question, albeit put slightly differently: if you'd been responsible for implementing this decision, what should you do before it's published?

I'll get my view on the decision itself out of the way first, because it'd be odd to pretend I don't have one. I think it was a bad call. Zachary Robinson's comment makes the principled case better than I could, and Julia Wise puts it more succinctly: 'don't hire someone with a fraud conviction for a finance or philanthropy role.' So I'll add only the practical one, which I don't think has been talked about: it's just not worth the hassle. Caroline Ellison is one person. Austin says he hired her because she was an excellent candidate, and I believe he believes that. But is she so exceptional that nobody else could have done the job? And is she worth all the ensuing very public drama and ostracisation? In my eyes, absolutely not.

But I work in operations, and in ops you regularly get handed decisions you'd never have made yourself. Your job then is to make them succeed, or at least soften their crash landing. So this post is an ops exercise: you're the Head of Ops or Chief of Staff at your org, your founder/CEO has made this call, and you get the Slack message about it. What do you do?

Alex, this exercise is silly, why are you bothering?

This might seem like putting lipstick on a pig. If the decision is wrong, a better rollout doesn't make it right, and I think that's true. Much of the fallout would've likely happened anyway, particularly because some folks believe a hard red line has been crossed (e.g. a recent fraud conviction would disqualify someone from a finance role at CEA).

But the aim of a good rollout isn't to neutralise the harm, it's to minimise it: to mitigate the risks to the organisation, its partners, the people who now have to explain to their own funders why they once used Manifund, and to Caroline herself. And I'd argue that the way the decision was rolled out also did a fair amount of damage, and Austin seems to half-agree. On 5 October he tweeted: 'perhaps I really botched the writeup.'

He also seems, a month on, to be doing several of the things below in reverse order: circulating a private letter to explain himself, running a poll, and considering a large personal donation to settle the $500k FTX lawsuit. Those are all reasonable moves. They'd just have been worth a lot more before the announcement than after it.

Finally, a caveat: this post isn't about how the decision to hire could've been better evidenced or reasoned (e.g. open round, modelling, strict criteria), but rather how the resulting implementation and announcement of that decision left a lot of value on the table.

Anyway, time to be a general after the war. Here's what I would've done.

Step 0: Disagree on the record, then set the conditions for reversal

Before doing anything, tell your CEO plainly and in writing why you think it's the wrong call. This is what disagree-and-commit is for, and it gives the decision-maker one real last chance to reconsider before anything is locked in. If things go tits up, your concerns will be documented.

Then, assuming your CEO is still set on it, I'd do a pre-mortem and agree in advance on the set of conditions that would trigger a reversal of the decision (or at least, further mitigating actions). For example, if specific partner organisations publicly withdraw, or if a set share of balances leaves the platform within three months. Deciding the threshold before the backlash starts matters, because once you're in the middle of the storm you at least have a break-glass plan, which lowers the cognitive load in the moment and makes it easier to overcome sunk costs.

Step 1: Keep the trial internal, then announce on day one

If a hire is controversial enough that you're tempted to give them a pseudonym, that's your signal. You wouldn't give any other new hire one, so the urge is telling you how people will react. The answer is to prepare for that reaction, not to delay it.

The legitimate worry behind a pseudonym is the one Caroline herself gave: she didn't want a public announcement during a work trial, only to be let go a month later. That's fair, and there's a better way to handle it. Scope the trial so that nobody outside the organisation needs to know who's doing it: internal projects only, no customer support, no published writing under the organisation's name. Trialists don't need to face the public anyway, and if you think that's necessary, you've underthought the trial.

Then, if the trial converts, announce on the person's first day as a staff member, alongside everything in Steps 2 to 4, not a month later. You could consider announcing a few days prior to their start date, but it's not immediately obvious to me why that'd be more beneficial (perhaps the pre-mortem exercise in Step 0 or the consultations in Step 3 would make that obvious).

Here's what happens otherwise. Caroline worked at Manifund as 'Carol' from her work trial on 13 July until the announcement on 11 September, including a full month after she went full-time on 10 August. She also published on the EA Forum under that name: Austin noted that 'Carol is the primary author' of a July post arguing that people shouldn't default to starting non-profits, which readers engaged with without knowing who wrote it.

Austin called this a 'minor deception.' I don't think it was minor. Every Manifund user, grantee and Forum reader who dealt with Carol was making a decision without the one fact they'd most want to know. People can only exercise agency when they have full information, and the pseudonym took that away from them. No wonder people are mad.

Step 2: Design the safeguards before you announce

Design the controls first, and publish them with the announcement. Not 'if you're worried, look for yourself', which is what Manifund's announcement offered: 'you (or your LLM) can take a look at Manifund's source code, underlying data, finances and research.' That's a do-your-own-due-diligence answer, and it puts the burden on exactly the people whose trust you're asking for.

It also undersold the problem. The one concrete example of Caroline's work in the announcement is a reconciliation tool that found misregistered transactions 'in the 5-6 figures.' This means she's working on the money ledger (or close to it), at an organisation with two full-time staff, so there's almost nobody to check her work.

Roughly, the controls I'd want in place:

• Separation of duties. The new hire doesn't approve, initiate or release payments, or change payment details, alone. At a two-person organisation, that means bringing in an outside party, like a board treasurer or an external bookkeeper, to be the second signature.
• Random review of their work. Periodic audits of code commits and any changes to financial records, by someone outside the organisation. In many health systems, a doctor returning to practice after addiction treatment agrees to random drug testing. It's undignified, but the point is safety, not comfort.
• Clear limits on the role. Write down what they can't do without sign-off, and publish it.
• Independent oversight. As user Bland Garlic noted, a competent independent board and someone checking the books are a low bar, and an announcement like this is the moment to show you clear it.

None of this will satisfy everyone. But it shows that the risk has been taken seriously by someone other than the people taking it.

I'd be remiss if I didn't also address the fact that Caroline Ellison is now your employee, and you'll have a certain degree of duty of care towards her as her employer. The above addresses how the organisation will be protected from her, but how do you intend to protect her from the public? Are you including her, to the extent that it'd be appropriate, in the comms planning and strategising? Does she have clear instructions about what her role is during the fallout period, whether she's expected to engage? Have you made adequate provision for mental health support? For all her crimes, Caroline Ellison is still a person, and she deserves to be treated like one.

Step 3: Tell the people who matter first, and give them weeks, not hours

As the ops lead, make a list of the partners, major donors and grantees whose reaction matters most, prioritise the top twenty or so, and speak to each one in confidence a few weeks ahead (ideally together with your founder/CEO). Not a forwarded draft; a conversation. Explain the decision, listen to the objections, take the anger, and change things where they have a point. (Some of their concerns could improve the safeguards in Step 2.)

This isn't about making people feel consulted. It's that people who hear something important at the same moment as the general public reasonably conclude that you didn't think they mattered, and they react accordingly. The ones who've trusted you most deserve to hear it from you first. Some will still walk away, but they'll do it privately and on better terms, rather than in a public comment thread.

This is just good change management: brief the people closest to it before you go wide, so the first time they hear it isn't the same moment everyone else does.

For comparison, as far as I can tell Manifund's advance notice consisted of forwarding the announcement on the Wednesday and Thursday before a Friday publication. Julia Wise says her team got an FYI about 16 hours before the post went up (Austin thinks it was more like 24). The date was set by Caroline being in town that week and Austin's son being due, by his own account.

One counterargument to this sort of stakeholder communication is that it risks a leak, and that as soon as you approach people and telling them it won't matter how many times you say 'in confidence', they'll get together and talk and it will leak, either by accident or otherwise. Admittedly, that could be worse because then you'd be on the back foot, but I'd keep two things in mind:

1. Many of them will have agreements with you that include confidentiality, and people generally honour a confidence they've been explicitly trusted with.
2. You can also be prepared for a leak scenario. You can have all your comms ready to go and any other materials you intend to publish or bilaterally send around.

Step 4: Write an announcement that only says what helps

Every line in the announcement should either explain the decision or reassure the people affected by it. Anything else gives critics material and helps nobody.

Manifund's 'How did I come to consider this at all?' section is a good example of what to cut. Point by point:

• Enjoying someone's Tumblr isn't a reason to hire them.
• Feeling 'a keen debt' to the FTX Future Fund reads, to almost everyone, as a debt to Caroline herself, which is no reason to hire someone and put an organisation at risk; it's become the centrepiece of both of the main rebuttals. Austin has since said that this isn't what he meant: the debt explains why he was willing to consider hiring someone with a conviction at all, when most people would rule it out on sight, and the decision itself came from interviewing and trialling her. That's a reasonable distinction, but if it takes a tweet four weeks later to make it, the original wording was doing more harm than good.
• Remaining conflicted on FTX is something you're entitled to feel, but not something to broadcast in a hiring announcement.
• Believing in redemption is a fine personal value, but the job of a chief executive isn't to give people second chances; it's to run an effective and efficient organisation. You're running a team, not a family.

Then add an FAQ section or page, because some of the questions are blindingly obvious. Why her? Was anyone else considered? What is she not allowed to do? What happens to the FTX lawsuit? It won't stop people asking, but it shows the decision has been thought through.

One important inclusion for the FAQs would be the BOTEC used to estimate whether the decision justified the costs (assuming one exists). A rough cost-benefit model does a lot of work. It sets the value of the hire's contribution against the expected cost in lost partners, lost donors and reputational damage to the wider movement, with honest uncertainty ranges. It's hard to model, but we're all EAs here and that's never stopped us before. Tellingly, in this case commenters like Roman Ross started doing the sums themselves within a day, using CEA's comms budget as a proxy for what the community will pay to protect its reputation. That's the analysis readers expected, and it should have come from Manifund.

Step 5: Get the tone right afterwards

When people tell you publicly that they're walking away, answer the thing they've said. Warmth that ignores it reads as not having listened.

When Aidan Alexander posted that AIM would no longer work with Manifund, Austin replied:

Thank you for the work you do at AIM; it's long been one of my favorite charities. At our wedding, my wife and I each picked a charity for attendees to donate to in lieu of a wedding gift; my wife picked The Humane League, and I picked the then-Charity Entrepreneurship, now AIM. AIM has also been one of the inspirations for the incubator I now organize, Surplus. Best wishes going forward.

He replied to EA Australia in much the same way. I'm sure it was meant warmly, but to someone telling you they're walking away, a story about your wedding reads as if you haven't heard them. Aidan wasn't asking whether Austin liked AIM.

A better reply would be three sentences:

1. I understand, and I'm sorry this puts you in a difficult position.
2. Here's what we're doing to address the concerns you've raised.
3. If there's something specific that would change your view in future, I'd like to know.

Austin's question

So, to answer what Austin asked: most of what I'd suggest now is what should have happened in August. Publish the safeguards. Publish the model, even a rough one. Get the lawsuit settled, which he's now considering. And keep talking privately to the people who walked away, without expecting them to come back soon.

While my view of Austin's judgement has severely downgraded, I think it's important to assume positive intent, which I think Austin held at the time and continues to hold. I hope I'm right about that, and I'm happy to be corrected on any of the above. I also hope the next organisation facing a decision this hard finds some guidance in my thoughts before the announcement, not after.

Acknowledgements

Huge thanks to Sophie Gulliver for her suggestions, review and feedback.