Ben Mann Monthly April 2022

Dog sitting, rafting, hikes, kayaking, flamingo, wars

Purpose

• An index for my memory

• A menu of topics for my next conversation with you

• A faster way to share what I’m excited about without the barrier of writing a complete blog entry on it

• A skimmable way to spread content I found valuable

Photos

Experiments and experiences

• Host doggie for 1w

  • Not all doggies are as nice as Roong-ji

• Coworking in Zurich and Zagreb

  • Working from Europe while most of your team is in California is great, as long as you don't mind working all day long. You can work undisturbed until the afternoon when the west coast starts waking up, then take your meetings until midnight. Better to cancel as many meetings as possible and take the focus bonus.

• Many hikes

  • Menaggio, Plitvice, Grindelwald, Lauterbrunnen

  • Mostly vertical with great views, repenting for my egregious ice cream and chocolate consumption.

• Rafting in Split

  • I'd never rafted before, but I especially enjoyed a brief detour through a cave and a pause by a 7m cliff jumping spot to search for a previous customer’s lost Apple Watch, which we found intact. Our guide was very good at teaching us how to raft, keeping us safe, helping us understand the local area and culture, and having a good time. Only the Brits drank the beer he brought for us 😅

• Fishing near Vis

  • Caught ~4 kg of John Dory, then went to a tiny island to barbecue them for lunch. Slept on the beach while the fish cooked. Pet some stray cats who were there to control the rat problem.

• Kayaking in Interlaken

  • Glassy lake, chill time, lots of sour gummies.

Life updates

• 🧳 Spent most of the month in Europe

• 📚 Started reading The New Fire and Cribsheet

• 📜 Published Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback

• 💸 Raised $600M to accelerate our research

Content

• Flamingo: Tackling multiple tasks with a single visual language model

  • Scaling up image/video/text multimodal models in a pretty simple architectural framework, impressive results.

  • Annoyingly vague on datasets

• Chris Blattman on the five reasons wars happen

  • Wars tend not to happen due to resource scarcity, but because of other pressures

  1. Unchecked interests — such as national leaders who bear few of the costs of launching a war.

  2. Intangible incentives — such as an intrinsic desire for revenge.

  3. Uncertainty — such as both sides underestimating each other’s resolve to fight.

  4. Commitment problems — such as the inability to credibly promise not to use your growing military might to attack others in future.

  5. Misperceptions — such as our inability to see the world through other people’s eyes.