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 711 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
Ratner: DataComp benchmark beat OpenAI models purely through data curation
“Just by cleaning, curating, sampling, filtering the data, we get a new state of the art score at compute parity, beating open AI models and others.”
Eifrem: Neo4j's 2021 funding round valued the company over $2B
“We raised this round, you know, last summer, and it's the first time that we went out with some numbers, like the valuation for the first time, which was north of north of two billion. And there's actually the largest round in database history, right?”
Craib: Numerai paid $10 million to data scientists in 2021
“We've even, I think even just last year we paid ten million dollars. So it's so much higher than the rewards are so much higher than the other data science competitions on online by orders of magnitude.”
Katz: ClickHouse will build a multi-tenant cloud-managed service
“And at the same time in parallel, we are going to build A multi-tenant managed service in the cloud, which will inevitably be deployed on a variety of different cloud platforms, whether it's AWS, GCP, Azure, whether we go to China, like I've done in the past a…”
Narayan: Timely Dataflow was first stream processor to match batch processing capabilities
“It was sort of the first, what I would describe as The very first stream processor that could do everything that batch processors could do.”
Running Apache Spark on a single node is slower than Pandas
“You can use spark at the single node scale as an alternative to pandas through the koalas interface, but you'll find that for many workloads, it's simply slower than pandas, which is not super impressive.”
Stewart: Amazon Aurora's single master limits write scaling and creates downtime risk
“If you're, if you have a lot of rights, they're still going to a single master. They didn't rewrite the execution engine. So if that master goes down, your apps offline, if you want to scale your rights, you're going to have to get a larger and larger master m…”
Pesenti: Facebook AI transformer models matched or beat Mathematica at math
“My team also applied transformer models to mathematics, you know, doing, like, things like partial derivation equation or integration and showing that, hey, it could work as well better than Mathematica or some, you know, hundred page long algorithm.”
The Cerebras Wafer Scale Engine is the world's largest processor
“This is the world's largest processor.”
Cerebras's WSE is the first chip with over 1 trillion transistors
“It is the first part that has over a trillion transistors, and we're here, and at 1.2, and it has 400,000 sparse linear algebra cores on it.”
Cerebras's WSE delivers 9 petabytes per second of memory bandwidth
“There's 18 gigabytes of memory right, right on that, ah, on that silicon wafer, and it gives an unprecedented nine gigabytes per second, sorry, nine petabytes per second of memory bandwidth.”
Cerebras routes around manufacturing defects across its 400,000 cores
“By having 400,000 independent cores, it's perfectly fine if we have a few hundred defects on there, and it's only our job to route the information flows around those defects.”
Cerebras's chip provides 10,000x the memory bandwidth of standard GPUs
“So, having solved this, we were left with a machine that was pretty impressive on most of the metrics, being larger, more cores, 10,000 times the memory bandwidth and 33,000 times the fabric bandwidth.”
Flux turns InfluxDB into a serverless execution platform for time series
“It essentially turns the database into a serverless execution platform for time series data. The idea is you can define any sort of custom logic that you want, inject it into the database, and it will periodically run that logic over the data that you're writi…”
Erb: Billions have been invested in pre-product AI hardware unicorns
“A few billion dollars have been invested in new hardware startups. There are a few unicorns already for companies that don't have products fully out yet.”
Lubin: Tech will soon allow querying data while keeping it encrypted
“In many cases very soon we're gonna be able to have data that remains encrypted at all times, yet we'll still be able to ask questions of that encrypted data”
Kulkarni: Elastic filed for IPO making $160M in revenue, more than MongoDB
“I think Elastic, which is Filed Furnace for a IPO, is an example of that. I think they're already making a hundred and sixty million In revenue, which is more than Mongo's making already, which is crazy.”
Stent: No generic text analytics platform can handle financial language
“There is no on the market generic text analytics platform that can handle this type of language. No way.”
Mostak: Tableau data extracts fail on datasets exceeding 100 million rows
“Tableau extracts are great, but they kind of fall over, either when the extract gets over a hundred million rows, a few hundred million rows, or when you have, like, streaming data doing the system.”
Golde: Uber drivers average 0.6 to 0.8 empty miles per passenger mile
“We look at some of our competitors, specifically Uber, there's a paper published again, either late last year or earlier this year, showing how many miles they ride empty on average for every single passenger mile that they do, and it's about between . Six and…”
Apple used Foursquare data to improve international maps after botched rollout
“Apple had a few problems with their map rollout, as you may have heard. And so you know, they came to us to try to improve the map data set around the world outside the U.S.”
Foursquare Predicted Chipotle's Q1 Sales Drop Within 0.3% Using Location Data
“We went on CNBC a couple weeks before Chipotle's Q-one earnings, and I said in two weeks they're gonna report their earnings, and you're gonna see same store sales down 30% in the United States post the E. Coli scale. And they were like, why should we trust yo…”
Bordes: Direct machine reading will eventually match IBM Watson's performance
“The goal is that, ah, eventually, by trying to understand the text directly, you can actually at least equal, all the information by Watson is in text, in free text.”
Corporate jargon in job postings disproportionately deters applicants of color
“Like some of that corporate jargon, the words like synergy and stakeholders and ROI. It turns out everybody hates them, but underrepresented groups, especially people of color, hate them even more. So nobody is as likely to apply when you include that language…”
WiredTiger storage engine made MongoDB ten times faster in version 3.0
“Wiretiger was not a little bit better than the previous storage engine. It really was a complete leapfrog. It was, you know, 10 times faster.”
Marcus: Autonomous driverless cars can drive in Palo Alto but not Manhattan
“In fact, they can drive in Palo Alto, but they can't drive in Manhattan.”
Marcus: Facebook M relies mostly on human operators rather than AI
“Facebook's new M service, which they have not rolled out at scale has humans on the back end. There's a little bit of AI in there, but it's mostly, ah, human beings, which is why they haven't rolled it out for a billion customers. They don't have enough human …”
Srivas: A single Apache HDFS cluster handles roughly 100 million files
“HDFS, a single cluster, can do about a hundred million files.”
Bloom: Netflix couldn't deploy its $1M prize algorithm due to complexity
“They paid a million dollar bounty, and they wound up looking at the code, and there were hundreds of separate models that were then boosted together, and they said there's no way we can do it.”
Bloom: 95% of production machine learning code is just glue code
“Algorithms are important, but 95% of all machine learning code in production is actually just glue code. It's connecting all of these different pieces together.”
Dextro was the first to offer automated video analysis as a service
“We were the first company to figure out how to get this level of analysis of what's happening in videos as a service.”
Stoica: Apache Spark won the TerraSort benchmark processing data out of memory
“Just October last year, we had this we won this kind of TerraSort benchmark. And in those, in that benchmark, the data, it's not in memory. Right? It's SSDs and so forth.”
Stoica: Apache Spark supports all major data workloads with one engine
“While we spark, You can use only one engine and only one API to support all these workloads.”
Bisignano: Most genetic research is based on European and Jewish populations
“Most of genetics actually is based on studying European and Jewish individuals just because that's been the body of literature that's been developed over the past few decades.”
LeCun predicted in 2014 that AI would mediate human social interactions
“And started thinking about the next 10 years. What are the next 10 years going to be for social interactions? And it's pretty obvious to a lot of people that a lot of our interactions you know, with our friends and a lot of interactions with the digital world …”
LeCun predicted in 2014 deep learning would conquer natural language processing
“There's a kind of a sense that in the community that the next set of techniques to kind of fall to deep learning, if you want, will be natural language processing.”
NOAA collects 20 terabytes of weather data daily but releases only 10%
“They still have 20 terabytes of data that they collect and create a day around weather, and they're only making 10% of that available.”
Global AIS ship tracking relies entirely on an unverified honor system
“So we have one hundred million data points a day analog, which are based on an honor system that trusts every vessel to tell whatever it wants about itself.”
One percent of global ships transmit false identities
“One percent of all the ships today transmit false identity.”
55% of global ships misreport their actual destinations throughout their journeys
“55% of ships misreport their actual port of call through all the journey.”
Gutierrez: Credit card availability to non-prime market dropped by $120B
“From 2009 to more recently, ah, there's been a reduction of about a hundred and twenty billion dollars of, Credit card availability to the subprime, nonprime market.”
Laplanche: Lending Club's expense ratio is under 2% vs banks' 5-7%
“That ratio for most banks is between five and seven percent, so it's really all operating costs divided by total loans outstanding, that's between five and seven percent. The same metric at Lending Club is less than two percent, and it's coming down quarter af…”
Firms spent $300M on fiber cables cutting NY-London latency by 5ms
“Ships now laying new fiber optic cables between New York and London to show five milliseconds of round trip time at the cost of three hundred million dollars.”
DARPA and intelligence agencies fund automated narrative manipulation research
“The KGB's just put money into this. IAPA and DARPA have got money going into this stuff.”
Sisense analyzed 10 terabytes in 10 seconds on a sub-$10k server
“We demonstrated Strata in Santa Clara last year, 10 terabytes, ok, analyzing 10 seconds on a sub-ten-thousand dollar standard Dell server.”