The Ledger
Every statement that passed quotation and attribution checks. Mix any filter with any other: certainty 1/5, debate potential 5/5, or both at once.
why aren't all 91 resolved? a statement only gets an assessment when the public record can support or contradict it. opinions and what-ifs never can, and 0 checkable ones are still open, waiting for their date. predictions held up or didn't; assertions are supported or contradicted. on every card: ▮▮▮▮▮ certainty · ▮▮▮▮▮ debate potential. speakers are clickable
Pesenti: Facebook uses embedding algorithms to automatically match misleading content
“And then when they fly content that should be at least, you know, shown as a misleading, then we have this very advanced similarity algorithm that kind of like look at an embedding of the of the content itself. Again, it can be the image itself or a multimodal…”
Pesenti: Facebook gained a year of usage growth in one month
“Basically the growth that happened in, just a matter of a month was what we predicted would happen over more than a year”
Borgman: Facebook runs essentially all of its infrastructure using ORC
“Essentially Facebook runs all of their infrastructure using ORC file.”
Tech giants like Google and Uber rely on home-rolled ML infrastructure
“Most of the tooling is kind of home rolled. Best practices haven't emerged yet. We're seeing a lot of, especially large companies, roll their entire own stacks. So Facebook has FB Learner, Google has TFX. Uber has Michelangelo. Airbnb has Big Head.”
Chris Dixon meets weekly with three teams leaving Google or Facebook
“I have three of these meetings a week where it's team coming out of Google or Facebook or someplace like that”
Glaser: Facebook feed algorithm generated historic business value in computing
“In fact, it would be hard to name a piece of machine intelligence that has generated more business value, maybe in the history of computing, than the rancor that the Facebook team built.”
Hoffman: Internal data represents under 0.01% of global data for most companies
“Most companies out there, their own data represents, like, point oh one percent of the world. It's, unless you're Google, Facebook, Amazon, Tencent, a couple of others, you have a very small sliver of what's happening in the world.”
Hoffman: Data overhead is the core struggle for non-big-tech AI companies
“So this is the course, the core struggle for almost every single company, except for maybe Google, Facebook, Amazon, Tencent, that has to deal with all these types of issues.”
Big tech uses AI and machine learning as its next feature war
“And three, and this is a belief of me that could be wrong, is that actually the big tech firms, Amazon, Facebook, Google, Microsoft, and all the others, they're using AI and machine learning as their next feature war.”
Production A/B testing at Yahoo and Facebook is a costly experiment
“If you think about A-B testing, that lives much more at the top of the ladder than at the bottom. You know, to actually get to the point where you're, If you're doing A-B testing in production at a large company like Yahoo or Facebook, you have to implement th…”
Bordes: Facebook AI Research is not tied to specific product groups
“And actually the lab is not even tied to any application or any product group.”
Srivas: Startups should avoid emulating Google or Facebook engineering infrastructure
“Technology wise, they are more advanced, but there's only one Google and one Facebook, right, and everybody else is not like that, so don't try to do that. I mean, as a startup, that's very extreme.”
In 2014, Google generated $30 per user annually versus Facebook's $6
“Google makes about 30 bucks a year per user. Facebook is doing a little bit worse. They're at six dollars per user”
Wilson: Three core differences define Twitter versus Facebook
“I think the important ones are default public versus default private follower versus friending. One-way follow model versus a two-way following model. And open fire hose versus closed fire hose. I think those three things tell you everything you need to know a…”
Budde: ZestFinance does not scrape Facebook data for credit decisions
“Everybody kind of, everybody first off assumes that we're scraping Facebook. We're not.”
Fernandez: Data startups can out-recruit tech giants by providing unique data
“Yeah, I think the diversity of skill set is critical, and the good news is if you're doing interesting things with data I know at least for us, like, I feel we have a shot at anybody, whether they're at Facebook or Google or doing, you know, wherever they're d…”
Stripe powers direct checkout inside Meta ads
“Meta is another example, a little bit of a different flavor, but we're powering checkout right inside ads.”
Zeghidour: Meta's LLaMA and DINO models were developed in Paris
“Lama was started in Paris. Dino, which is the most groundbreaking vision work from Facebook was developed in Paris.”
Patel: Meta operates two MTIA chip lines for recommendations and Gen AI
“Meta actually has two lines of AI chips. They're MTIA. There's a line that's focused on recommendation systems, and then there's a line that's focused on Gen AI.”
Lin Qiao: Meta had hundreds of engineers building PyTorch and its infrastructure
“We have hundreds of engineers building PyTorch and infrastructure around PyTorch, but at the same time, I believe PyTorch within Meta probably has thousands of users.”
Meta spent five years rebuilding PyTorch's backend for internal scale
“It took us five years. Took us five years to get the stage supporting almost all internal needs using deep learning and mass and massive scale.”
Masad was among the first four engineers on React Native
“So this project called Catalyst at the time, which became React Native, and I was the first to I was one of the first, like, three or four people to work on it.”
Benaich: Meta reported a 7% to 10% click-through gain on generative ads
“He gave some stats around how consumers click through at a higher rate, and it was material, it was like something in the range of seven to 10% click through improvement on generative ads.”
Van Luijt: OpenAI recommended Weaviate when deprecating its search endpoint
“OpenAI had a they had a search endpoint that they deprecated. This is pre-ChatGBT, so, but they, so they deprecated that search endpoint, And say, well, you know, you just can buy our embeddings. There are three ways that you can index them and search for them…”
Falcon: Lightning AI hired core Meta PyTorch leads
“We have an in-house PyTorch team, so we hired a lot of the core leads from Meta”
Runway's enterprise clients include CBS, Google, Facebook, and Vox
“We help companies like CBS, Assembly, Google, Red Lab, Facebook, VaynerMedia, Vox, RGA, and countless other companies create better video content faster and literally at a fraction of the cost they're creating it today.”
Pesenti: Facebook has more ML engineers outside core AI team than inside
“And actually there are more ML engineers outside my team than within.”
Pesenti: Facebook has as many data scientists as product managers
“So actually, you know, to get an idea, there are as many data scientists in Facebook as they are product managers, right?”
Pesenti: Facebook trains single XLM-R models across 100 languages simultaneously
“We try to learn a hundred languages at the same time in a single model using transformer architecture”
Pesenti: Facebook is not an AI vendor and has no cloud offering
“So for us, actually, we're not selling AI. We're not a vendor. We don't have a cloud offering.”
Pesenti: FAIR lab had about 300 researchers across eight offices in 2020
“We don't share, but you're yes. Bullpark is good. Yeah.”
Borgman: Facebook Runs the Largest MySQL Deployment
“Facebook happens to run the largest MySQL deployment as well.”
Goldman Sachs' internal surveillance graph has 100M nodes and 1B edges
“Or in our case, slightly smaller than the Facebook graph, but still about a hundred million nodes and about a billion edges.”
Vision, language, reasoning, and planning form Facebook's ten-year AI roadmap
“Understanding vision, understanding language, reasoning, and planning, I didn't even talk about planning, are actually a very key element that fits in the whole Facebook Facebook strategy for the tenures.”
Neustar maintained data partnerships with Facebook and Twitter for predictive analytics
“We have a partnership with Facebook. We have a partnership with Twitter also. We use their data, ah, to be able to run our analysis as such.”
Facebook data infrastructure capacity planning consistently lagged behind demand
“Having, you know, run big clusters at Facebook, I remember conversations with the procurement team where it was like a big capacity planning exercise trying to predict the future, and yes, I know predictive analytics works to predict the future really well, bu…”
MySpace in 2007 enjoyed the market dominance Facebook has today
“That was back in 2007 where MySpace, you know, had kind of was at the height where probably Facebook is today.”
Mason: Link half-life is 2.8 hours on Twitter versus 3.1 on Facebook
“We found that for Twitter, it's short, it's 2.8 hours. For Facebook, it's a little longer, about 3.1 hours.”
Masad broke Facebook's Android beta for 10 million users during bootcamp
“At some point I broke the Android app for like ten million ten million users, the beta users”
Matt Turck: Ada has over 300 customers including Meta, Verizon, and Shopify
“Today, the company has over 300 customers using the platform, including Meta, Verizon, and Shopify.”
Ashish Thusoo led Facebook's data infrastructure team before founding Qubole
“Before I started QBOL, I was at Facebook, and I built out, I ran the team that built out the infrastructure there.”