Some early lessons from the world of work

A scattered and lengthy collection of thoughts I might come back to tidy up in future.

External communications

• It’s worth meeting people where they’re at, especially if they’re doing you a favour by chatting (e.g. informal advisors, references)
  • Tech people love Signal with disappearing messages
  • When responding to an intro, include your scheduling link but also pre-emptively say you can offer alternative times if needed
• Having someone’s number in your phone is often unexpectedly helpful, e.g. if you subsequently need to ask for a small favour at short notice (it feels much lower friction for them than responding to an email)
• Gemini meeting notes is so helpful
  • And when I asked ~15 people whether they minded me using it, many seemed surprised that I was even asking. (but I think still good to ask)
• I think I’m pretty good at politely & kindly sending messages that might be disappointing for the recipient, and this helps me save lots of time & energy when responding to emails, meeting requests, etc
  • This took a while to get comfortable with.
  • Rejecting people is hard, and it’s very tempting to say nothing substantive or just be wildly inaccurate in emails turning people down.
    • This isn’t helpful. Outlining your reasons for rejecting someone can be done fairly easily, seems better than patronising them, and if they reply trying to rebut you can just ignore it in the worst case.

• I’m a lot better at writing emails than any current LLM is, especially when you take into account my time
  • Potentially with access to my whole email corpus they could pick up on the correct style, but a lot of email writing requires loads of additional context on the state of the business relationship to determine what tone is best.
    • I’m still working on getting better at this.
• When interfacing with Bay Area (& generally US) people, I can often be less apologetic & more direct than my British default would be
• It’s good when the subject line is a question. You can also append the org name if relevant, e.g. “Investment in Acme Seed Round? / Sonicspeed Ventures”
• Make it easy for people to forward your email on to the relevant person, if asking for an intro.
  • Sometimes, this means adding in more context than is necessary or feels natural.
  • See this for an explanation
• Keep messages brief. Use bullet points and bolding. If you need a quick response, make sure that the ask is simple (and prepend “[Time-sensitive] to the subject line”).
  • On rare occasions I think longer emails can be better. In particular, if you’re dealing with bureaucracy and want to convince somebody that you’ve given lots of thought to whatever rule/topic/requirement it is they care about, having a very verbose message can serve like a kind of “proof of work” which wins them over.
    • Even then, though, make sure that the first paragraph communicates the crucial point! It’s unlikely that they’ll actually be so careful as to read the whole message; I think it’s mostly the appearance of effort on your part which matters.

Feedback, management, and hiring

• Saying the obvious, but I find specific, positive feedback very motivating!
• The project of improving at being a worker is quite a fun one to me, I think partly because it involves so many complex (social) skills.
  • For example, I find the idea of becoming a top 1% manager exciting, and would be keen to put in the work to get there.
  • Especially as I can offload more cognitive work to LLMs, the interpersonal aspects of work feel like one area where development will continue to be very meaningful & satisfying.
  • I make careful notes of constructive criticism I’ve received; I think it’s quite good to orient towards them as precious gems
• The frustration of others (seeming to) steal credit is real.
  • It’s difficult to let go, because things just don’t seem fair. Even if you’re bought into the mission and mostly care about the outcome, the idea that the credit-stealer is still out there receiving status for your work really grates.
• If an org lacks a culture of everybody enthusiastically giving & receiving candid feedback, there’s a risk that performance issues will not be raised until it’s too late.
  • If you wait for problems to develop until breaking point, then you’ll either have to send a big list of grievances – which doesn’t set the stage well at all for a productive conversation later – or only address a subset of them.
  • In general, I think it’s better to discuss feedback informally in-person first.
  • I’m still working on getting better attuned to how much feedback to give in different circumstances, in order to get the best expected outcomes. It’s easy to feel like you’ve been too harsh and then overcorrect.

• The Clearbit guide on management is excellent across the board
  • I especially endorse the sections on impeccable agreements and running meetings
  • Having at-least-fortnightly 1:1s with coworkers has been very useful for keeping people on track & in sync
• There are lots of loose ends for handovers, and missing out on things makes life very difficult for whoever is taking over
  • Yet another reason to use a password manager and explicitly document processes!
  • Email threads are the worst for handovers, because they often contain lots of context but aren’t easy to share with others
    However, some of the permissions that Google Workspace super-admins have are pretty surprising. They can transfer all your Google Drive files, and see details of your calendar events if you share availability with colleagues.
    .
  • Joey Savoie gave a great EAG talk about what makes CEO transitions go well; hopefully it’ll be on YouTube soon.
• Volunteer management is difficult.
  • You’re very reliant on mutual goodwill. Setting firm expectations about responsibilities / deliverables doesn’t really work, because if someone is not doing what’s been agreed there just aren’t many consequences you can credibly commit to.

• Even short screening calls can be very high-value.
  • Often you care less about gathering facts and more about seeing how somebody thinks.
    • It hadn’t really occurred to me that e.g. in a call with a founder about their startup, you might want to evaluate their general reasoning & judgement, as opposed to get further info about business metrics – even though in the context of interviews for hiring (or, e.g. university admissions!) it’s pretty obvious that this is one of the primary goals.
  • You might also be interested in how someone presents information to you, and thus not try too hard to ask for specifics in order to reduce their scope to manoeuvre.
    • The choices people make in how they approach open questions can be informative: how do they decide to spin things?
• The rule of thumb that your total number of applications doubles over the closing weekend seems roughly correct.
  • This is one reason to put in a deadline even if you don’t particularly need it.
  • Obviously setting deadlines means you potentially miss out on great applicants who heard about the posting too late.
    • One mitigation here is to continue accepting applications after an initial soft deadline; I’m not sure if this has a strong enough psychological incentive to push people into applying though.
    • Unfortunately, I think even people who turn out to be excellent candidates sometimes procrastinate applying. So using the lack of a cut-off date as a deliberate selection mechanism doesn’t really work well.

Focus and efficiency

• It’s easy to get distracted from important deep work by urgent tasks that ultimately don’t matter much
  • This problem is especially acute when working part-time, because it’s hard to get long blocks of focus time, so you’re pushed even further towards only doing the itty-bitty work.
  • In addition, I find that it takes a certain amount of effort to «warm myself up» into doing cognitively difficult tasks.
    • So, I prefer not to do them when I only have, say, one hour free and thus won’t get to reach the satisfying-but-challenging flow state.
• Improving processes within your org can often be high-leverage.
  • For example, setting up a better spending tracking system might not only save your ops lead time, but also dramatically lower the barrier for other colleagues to look at the data and thus make better decisions
• I’ve found it really helpful to systematise things with template docs / SOPs.
  • Relatedly, I would find it impossible to manage without Todoist or some equivalent – having to worry about tasks I’m only keeping track of in my head would make things super stressful.
• Time is the scarcest resource. Use all the LLMs, to the extent where you can figure out which will be best-suited for a particular task
  Amusingly, I ran this blogpost through o3 asking it “What does the author get wrong?”, and it responded with very comprehensive catalogue of quotations accompanied with wildly overconfident & entirely hallucinated evidence supposedly contradicting my points. It is indeed a lying liar.
  (and e.g. can identify which model produced an output by its tone & quirks)
• Some people will have ineffective executive assistants you have to interact with, and this makes everybody’s lives harder

Internal collaboration

• Avoid information overload!
  • Doing the following makes it more manageable to stay on top of what matters:
    • Bold important phrases / sentences for people to skim-read
    • Hide link previews and avoid lengthy top-level posts (instead put details into a thread)
    • Don’t over-use “send to channel” in Slack
• If you’re writing up a doc for your manager, present a recommended option with reasoning, and then some alternatives.
  • Don’t make them decide all by themself; they should ideally be able to reply in a few minutes.
  • Also, to avoid wasting people’s time, it’s best to specify very clearly the kind of feedback / input you want.
    • “Tell me if there’s anything obviously missing from this plan for how I’ll research X” is very different from “Carefully read through this section and tell me if there are any errors in reasoning”, and also different from “Help me reword the doc into a tone that you think sounds better”.
    • Sometimes people default to the last of these types, if not explicitly asked for something else.
• If something can happen in a Google Doc
  Personally, I’d prefer to use the Office suite, but everybody in the VC/AI/EA world seems to be with Google Workspace.
  , have it happen there.
  • In particular, I find that it’s an extremely poor use of time to discuss high-level ideas/strategy in meetings, both because it takes longer and your discussion will be lower-quality than if it were done through written communication
  • Also, the most loquacious people will dominate the conversation, whereas with written discussion you can skim-read and choose where to direct your attention.
• Common knowledge about how quickly to expect a response from people is very important
  • I think an expectation of same-day responses (during the week) is reasonable. Have an #urgent channel if you might need it.
  • I much prefer being my own bottleneck than being in a situation where I’m blocked by others, but in circumstances where that’s not possible, it’s at least good to know how long I’ll have to wait
• You become acutely aware of the flaws of each communication medium.
  • It’s frustrating that you can’t directly reply to a message in Slack, and need to use the inline quote formatting instead
  • On the other hand, the ability to copy a link to a message in Slack is very useful; this is one reason I prefer having conversations there vs Signal or email.
  • I’m glad Google Docs comments allow for bolding & italicising, but the lack of bullet point support is wearying. It’s also sometimes irritating that, unlike with Word, you can’t really copy comments between documents or delete other people’s comments.

Compliance

• If you think doing a self-assessment is complicated and annoying, just have a try of being a legal (rather than natural) person!
• Regulatory considerations are a real constraint on possible actions and can quite heavily shape what business decide to do.
  • I’m not just talking about trivial things like “Aw shucks, I guess we can’t pay employees less than the minimum wage
    Although this itself is surprisingly complicated.
    ” – compliance considerations come into play when looking at entering a new business area, or supporting an ambitious project, or making an otherwise-profitable trade, and so on.

    

• Toy example: guess where the UK VAT threshold is, using the graph above.
  • Now think how every business there has made a conscious decision not to grow more, and needs to actively weigh up the costs vs benefits of crossing the threshold with each potential contract.
• Unfortunately, there do have to be thresholds for these sorts of compliance things.
  • Whereas for the UK childcare tax trap, you could taper the allowance and easily solve the problem, there’s not really a way to half-do things like independent audits.
  • So you will end inevitably up with discrete changes in requirements, and perverse incentives.
• Tax codes are extraordinarily complicated.
  • e.g., see the guidance about UBTI and ECI: there are so many exceptions and counter-exceptions designed to close off certain loopholes while no doubt also making carve-outs for particularly noisy lobbies.
  • Or have a go at figuring out the differences between B-, C-, S-Corps, LLCs, LLPs, and LPs (!)
• I think it’s quite useful trying to understand why Congress passed these rules, as a way of building up an intuitive mental model of how they might affect you.
  • Pre-LLMs, it makes sense that lawyers were like wizards: they’re able to give you tailored guidance about what is & isn’t relevant, saving you from the impossible task of parsing & digesting all of the legal code.
  • And even with our current LLMs, input from experienced counsel is still pretty valuable, especially when you’re in a slightly unusual situation.
• Following the example of an established org (or recommendations from an experienced individual) is really helpful in this regard: they’ll have a very useful map of the non-obvious considerations that will turn into gotchas if you’re not careful.
  • It would be great if more of this information were easily accessible and reliable; it’s really difficult to judge the quality of internet “How to do X” listicles without an endorsement from someone you trust.
  • Here’s one excellent-seeming guide on starting a US nonprofit.

See also

• Peter Wildeford’s review of “Managing to Change the World”
• Daniel Kestenholz’s notes on management
• Various posts from Ben Kuhn
• Lincoln Quirk on organisational culture