Andrew ‘Boz’ Bosworth: evaluation

Transcript context

…Okay. So you talked about the newsfeed, which was one of your very earliest projects at Facebook. Here's a couple of fun facts I know about the newsfeed. One is that it was the very first algorithmic newsfeed of its kind of any social network and maybe of any sort of product like this. Yeah. And two is the very first AI code that was written in Facebook to rank the actual newsfeed. So there were a lot of firsts and clearly this became a huge deal. The newsfeed is essentially what people think of when they think of Facebook now. But it was super controversial when it came out. People were very against this. They did not want to be sharing this much information with people, or so they thought. And then they realized eventually, oh, this is actually exactly what I want. What did you learn from going through that experience of building something that people initially reject and then later realize that they actually do want this and this is exactly what they're waiting for? This is a story that you tell a lot actually in through your interviews, which is just like you have to have conviction in what you're building. You're choosing your customers as much as your customers are choosing you is one way I think about it sometimes. And one mistake that you do see sometimes startups make is they get an early cohort of users whose needs actually take them kind of orthogonal to a larger market. And so they become kind of held hostage by their earliest customers. Now, so time and time again, we've had a vision for what we thought this should look like. And it wasn't the thing we were delivering right now. And so people who are using the thing we currently had, we're not sure that that changes what they wanted. But we had a confidence that over time they would. And we're not always right. But in these cases, we were right. Now, News Feed was an easier case than people suspect because everyone was outraged at the same time as they immediately doubled their usage of the product. So we had a few advantages there, which was there was literally like everyone was like, I hate this so much. And they would refresh, refresh, refresh. And so we're like, OK, wait, this is a. There's cognitive dissonance here between what the stated preferences and what the revealed preferences are in the economic sense. So News Feed was a little easier than people suspect to stick with. But people sometimes misunderstand that. They think, oh, the lesson is don't listen to your customers. Not at all. And we certainly care tremendous about, and even with News Feed, we did actually screw some things up. I kind of always make this joke that it's almost like, you know, you're in here at the party and music's loud, you're talking to somebody, and the music cuts out right when you're saying something at a super high volume. And so everyone in the party hears the last thing you said. Now, you were saying it in a public place, so it wasn't like it was a private comment, but you also didn't mean to broadcast it at that volume. We kind of did that to the entire user base because we took what had been wall posts, which, sure, anybody could have gone to that profile and seen and then put it kind of on blast, like on main, you know, as the kids say these days, put it on main. And someone's like, ah, you know, so we did that. We I do think we like I don't want to say like we we did screw things up. Like it wasn't like, oh, this is a clause execution. So another thing to know is, when did you screw something small up? And when did you screw something big up? When is the thing itself wrong versus when were the details wrong? That is an art. That is a real art. And you don't always have user data to determine it like we did. And so a lot of that is, do you have a clear vision and intuition for what you expected to happen, and then what happened instead, and can you diagnose the delta there? So the newsfeed case, we made a bunch of little mistakes. o you have a clear vision and intuition for what you expected to happen, and then what happened instead, and can you diagnose the delta there? So the newsfeed case, we made a bunch of little mistakes. The thing itself was right. And I really am quite proud of the work we did there. You know, me and Chris Cox at the most core, probably in the engineering side, which you saw on the PM, like there was no ranked feeds before that. We did have some AI that I'd built before for the anti-spam, anti-things, but it was pretty rudimentary. So it was probably the first consumer AI that was in a website of that kind around content. And we built like the most efficient monetizing surface in history outside of search, I think. And for those who are curious, I don't use monetization because I think money is the most important thing. I do think it suggests the economic power you've created, which I do think correlates very strongly with human utility. Although, you know, obviously I respect that some people may disagree.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.