Ever wondered what kids are really doing online?
Qustodio aggregated screen time data from 400k+ kids aged 4-18,the results are fascinating.
Roblox dominated screen time across categories, with players engaged for 3 hours/DAY. Almost 60% of kids in the sample played Roblox, followed by a distant #2 for games in Clash Royale (24%) and #3 in Minecraft (23%).
YouTube is still in kids’ video streaming, followed by Netflix.All other major services (including Disney+) have seen a decline in screen time over the last 3 years.
Instagram had a particularly tough year with kids. It fell out of the top five social media apps (by % of kids in each country who use it) in the US, UK, and Australia.In the US - Reddit and Twitter now both rank higher.
32% of kids globally are now on Discord - it’s #2 in messaging behind WhatsApp.
Education apps see the lowest time spent across the major categories.
Duolingo is the clear global winner in terms of percentage of devices installed, followed by Photomath.
Full report here.
Qustodio aggregated screen time data from 400k+ kids aged 4-18,the results are fascinating.
Roblox dominated screen time across categories, with players engaged for 3 hours/DAY. Almost 60% of kids in the sample played Roblox, followed by a distant #2 for games in Clash Royale (24%) and #3 in Minecraft (23%).
YouTube is still in kids’ video streaming, followed by Netflix.All other major services (including Disney+) have seen a decline in screen time over the last 3 years.
Instagram had a particularly tough year with kids. It fell out of the top five social media apps (by % of kids in each country who use it) in the US, UK, and Australia.In the US - Reddit and Twitter now both rank higher.
32% of kids globally are now on Discord - it’s #2 in messaging behind WhatsApp.
Education apps see the lowest time spent across the major categories.
Duolingo is the clear global winner in terms of percentage of devices installed, followed by Photomath.
Full report here.
Samsung, Google and Qualcomm are joining hands to make a mixed-reality platform.
These companies only announced the partnership — they had no future device or hardware product to show.
These companies only announced the partnership — they had no future device or hardware product to show.
TechCrunch
Samsung, Google and Qualcomm are making a mixed-reality platform | TechCrunch
Samsung, Google, and Qualcomm are joining hands to make a mixed-reality platform, the company said at the Galaxy Unpacked event.
MakerDAO just announced Spark Protocol: a competitor to Aave.
It's a sharp turn from Maker's old benign strategy of integrating DAI into existing DeFi protocols. Now Maker wants to build its own DeFi ecosystem.
Maker's strategy has been focused on integrating $DAI into leading DeFi protocols.
For example, Maker recently enabled Dai Savings Rate (DSR) with a 1% APY yield.
The expectation was that Aave or Compound would integrate DSR and USDC holders would move to DAI.
That hasn't happened.
Instead, Maker is forking Aave v3 with DSR integration and EtherDAI at the center.
Sure, Aave gets 10% of the revenue, but the message is clear: Thanks boys, we are going to do it ourselves now.
It's no surprise when you think about it.
Aave is about to launch its own DeFi stablecoin GHO—a clear competitor to DAI.
Curve is also launching a stablecoin crvUSD, and Curve is crucial for DAI's liquidity on-chain.
What's more, Maker is soon launching EtherDAI.
Similarly to Frax's frxETH, EtherDAI is a synthetic liquid staking derivative (LSD) for ETH.
It's a core part of Maker's Endgame plan to make DAI the Unbiased World Currency.
That's not all.
Maker's Spark Protocol will support two oracles: Chronicle 3 (prev. Maker Oracles) and Chainlink.
It's an extra security step in case on oracle fails.
Spark Protocol will have its own token, but it will be wholly owned by Maker's.
Maker's Endgame Plan envisions creating MetaDAOs (or SubDAOs) to reduce the complexity of how Maker is governed.
You can think of Maker Core as L1 Ethereum: slow and expensive, but secure.
While MetaDAOs are L2 solutions: fast & nimble, that get security from the L1.
Spark Protocol will be the fast & nimble part of Maker.
Each MetaDAO has a specific task at hand.
Spark Protocol will serve as the backbone lending protocol for all current and future Maker's services, like DAI or EtherDAI.
It's like Google's Alphabet: Maker Core is Google, and MetaDAOs are the “Alpha Bets”.
Each MetaDAO will have a token launched via yield farming - this is how we'll get Spark Protocol token.
No pre-allocation or private/public sales. A real feast for yield farming thirsty degens.
SPOILER ALERT: You’ll need MKR to farm every single SubDAO token.
It's a sharp turn from Maker's old benign strategy of integrating DAI into existing DeFi protocols. Now Maker wants to build its own DeFi ecosystem.
Maker's strategy has been focused on integrating $DAI into leading DeFi protocols.
For example, Maker recently enabled Dai Savings Rate (DSR) with a 1% APY yield.
The expectation was that Aave or Compound would integrate DSR and USDC holders would move to DAI.
That hasn't happened.
Instead, Maker is forking Aave v3 with DSR integration and EtherDAI at the center.
Sure, Aave gets 10% of the revenue, but the message is clear: Thanks boys, we are going to do it ourselves now.
It's no surprise when you think about it.
Aave is about to launch its own DeFi stablecoin GHO—a clear competitor to DAI.
Curve is also launching a stablecoin crvUSD, and Curve is crucial for DAI's liquidity on-chain.
What's more, Maker is soon launching EtherDAI.
Similarly to Frax's frxETH, EtherDAI is a synthetic liquid staking derivative (LSD) for ETH.
It's a core part of Maker's Endgame plan to make DAI the Unbiased World Currency.
That's not all.
Maker's Spark Protocol will support two oracles: Chronicle 3 (prev. Maker Oracles) and Chainlink.
It's an extra security step in case on oracle fails.
Spark Protocol will have its own token, but it will be wholly owned by Maker's.
Maker's Endgame Plan envisions creating MetaDAOs (or SubDAOs) to reduce the complexity of how Maker is governed.
You can think of Maker Core as L1 Ethereum: slow and expensive, but secure.
While MetaDAOs are L2 solutions: fast & nimble, that get security from the L1.
Spark Protocol will be the fast & nimble part of Maker.
Each MetaDAO has a specific task at hand.
Spark Protocol will serve as the backbone lending protocol for all current and future Maker's services, like DAI or EtherDAI.
It's like Google's Alphabet: Maker Core is Google, and MetaDAOs are the “Alpha Bets”.
Each MetaDAO will have a token launched via yield farming - this is how we'll get Spark Protocol token.
No pre-allocation or private/public sales. A real feast for yield farming thirsty degens.
SPOILER ALERT: You’ll need MKR to farm every single SubDAO token.
Twitter
I’m super excited to announce the launch of ⚡️ Spark Protocol ⚡️ - a growth-focused arm of the Maker Protocol.
Our first product is Spark Lend which builds upon Aave V3 to bring the best lending market features to Maker.
🧵👇
Our first product is Spark Lend which builds upon Aave V3 to bring the best lending market features to Maker.
🧵👇
Tether reports $700 million Q4 net profit in latest attestation report — link
The Block
Tether reports $700 million Q4 net profit in latest attestation report
Stablecoin issuer Tether released its latest attestation report on Thursday, reporting a profit of $700 million in the fourth quarter of 2022.
First-ever legal trial on NFTs – "NFTs aren't art"
Hermès won its lawsuit against the digital artist behind “MetaBirkin” NFTs.
The jury found that Rothschild had infringed on Hermès trademark rights and awarded the company $133k in damages, arguing that the NFTs were similar to commodities and not protected by the First Amendment as art.
This ruling could set a precedent for other big brands seeking to protect their trademarks in the NFT
Hermès won its lawsuit against the digital artist behind “MetaBirkin” NFTs.
The jury found that Rothschild had infringed on Hermès trademark rights and awarded the company $133k in damages, arguing that the NFTs were similar to commodities and not protected by the First Amendment as art.
This ruling could set a precedent for other big brands seeking to protect their trademarks in the NFT
NY Times
Hermès Wins MetaBirkins Lawsuit; Jurors Not Convinced NFTs Are Art
The verdict means First Amendment protections did not apply to a digital project blurring the line between art and business.
eCommerce is the biggest Metaverse
All Metaverse market segments are experiencing immense growth.
The entire Metaverse market is expected to surge by an impressive 759% between 2022 and 2027, shooting up from US$39 billion to a worth of US$335 billion.
eCommerce is currently the biggest Metaverse market segment with a volume of US$16 billion.
Although it is going to lose some of its market share within the Metaverse market in the future, eCommerce is expected to stay as one of the most important Metaverse pillars through 2030.
A growth of 375% secures the eCommerce segment the top spot in the market through 2027, with a market volume of US$76 billion.
By 2030, Virtual Assets will overtake eCommerce as the biggest Metaverse market segment. In general, the Metaverse opens up a huge world of new opportunities for brands and retailers to attract and satisfy customers.
All Metaverse market segments are experiencing immense growth.
The entire Metaverse market is expected to surge by an impressive 759% between 2022 and 2027, shooting up from US$39 billion to a worth of US$335 billion.
eCommerce is currently the biggest Metaverse market segment with a volume of US$16 billion.
Although it is going to lose some of its market share within the Metaverse market in the future, eCommerce is expected to stay as one of the most important Metaverse pillars through 2030.
A growth of 375% secures the eCommerce segment the top spot in the market through 2027, with a market volume of US$76 billion.
By 2030, Virtual Assets will overtake eCommerce as the biggest Metaverse market segment. In general, the Metaverse opens up a huge world of new opportunities for brands and retailers to attract and satisfy customers.
Ecommercedb
eCommerce is the Biggest Metaverse | ecommerceDB.com
<p>By 2030, eCommerce in the Metaverse is expected to reach a market value of US$191bn, up from US16bn in 2022</p>
The AI Chatbots war is on
- Microsoft: ChatGPT
- Google: LamBDA
- YouSearchEngine: YouChat
- Baidu: ErnieBot
- Perplexity AI: Perplexity Chat
- AnthropicAI: Claude (GOOGLE)
- Jasperai: Jasper Chat
- Microsoft: ChatGPT
- Google: LamBDA
- YouSearchEngine: YouChat
- Baidu: ErnieBot
- Perplexity AI: Perplexity Chat
- AnthropicAI: Claude (GOOGLE)
- Jasperai: Jasper Chat
Theory of Mind emerged in GPT
Children get it at 9yo
No, it doesn't read minds (yet) but,
It can empathize and imagine what you know
separately from what it or others knows
Children get it at 9yo
No, it doesn't read minds (yet) but,
It can empathize and imagine what you know
separately from what it or others knows
👍6
Microsoft Kills Its Industrial Metaverse Team After 4 Months
In a surprising reversal,Microsoft has killed a team it formed four months ago to help customers use the metaverse in industrial settings.
The group’s roughly 100 employees have all been laid off.
Microsoft created its Industrial Metaverse Core team in October to help build software interfaces for operating the control systems behind electrical power plants, industrial robotics and transportation networks.
The team, primarily composed of a group known internally as Project Bonsai, fell victim to Microsoft’s plan announced last month to lay off 10,000 employees, or about 4.5% of its workforce.
In a surprising reversal,Microsoft has killed a team it formed four months ago to help customers use the metaverse in industrial settings.
The group’s roughly 100 employees have all been laid off.
Microsoft created its Industrial Metaverse Core team in October to help build software interfaces for operating the control systems behind electrical power plants, industrial robotics and transportation networks.
The team, primarily composed of a group known internally as Project Bonsai, fell victim to Microsoft’s plan announced last month to lay off 10,000 employees, or about 4.5% of its workforce.
The Information
Microsoft Kills Its Industrial Metaverse Team After 4 Months
In a surprising reversal, Microsoft has killed a team it formed four months ago to help customers use the metaverse in industrial settings, according to a person with direct knowledge of the matter. The group’s roughly 100 employees have all been laid off…
Crawling the Internal Knowledge-Base of LLMs
- Given a seed entity, expand knowledge-graph around it
- Crawling decomposed into sub-tasks, w/ prompts that control for both:
precision (no wrong facts generated) &
recall (number of facts generated)
Paper
- Given a seed entity, expand knowledge-graph around it
- Crawling decomposed into sub-tasks, w/ prompts that control for both:
precision (no wrong facts generated) &
recall (number of facts generated)
Paper
👍1
Data decoding and Deserialization issues can easily became the #1 API security risk for 2023. Log4j, spring4shell, Text4Shell are just a few remarkable, among hundreds of others.
GitHub
CVE-2023-25194 - GitHub Advisory Database
Apache Kafka Connect vulnerable to Deserialization of Untrusted Data
PayPal held $291 million of Bitcoin and $250 million of Ether — link
CoinDesk
PayPal Held $604M of Customers' Crypto as of Year-End 2022
The payments company held $291 million of bitcoin and $250 million of ether, with the rest comprising litecoin and bitcoin cash.
"Are Deep Neural Networks SMARTer than Second Graders?"
Experiments reveal that while powerful deep models offer reasonable performances on puzzles that they are trained on, they are not better than random accuracy when.
LLMs (and big curves in general) can store and reuse human-generated abstractions that they're exposed to, but cannot generate their own abstractions when faced with a new problem.
Fundamentally, this is the crux of the issue. We see systems that rehash human output, and we're keen to attribute human abilities to them. But what makes humans intelligent is that they *invented* this stuff in the first place. Humans aren't sponges, they're creators.
Experiments reveal that while powerful deep models offer reasonable performances on puzzles that they are trained on, they are not better than random accuracy when.
LLMs (and big curves in general) can store and reuse human-generated abstractions that they're exposed to, but cannot generate their own abstractions when faced with a new problem.
Fundamentally, this is the crux of the issue. We see systems that rehash human output, and we're keen to attribute human abilities to them. But what makes humans intelligent is that they *invented* this stuff in the first place. Humans aren't sponges, they're creators.
Restrictive crypto rules for EU banks confirmed in published legal draft — link
CoinDesk
Restrictive Crypto Rules for EU Banks Confirmed in Published Legal Draft
Banks in the European Union would have to treat crypto as the riskiest kind of asset and disclose exposures while awaiting more detailed rules.
HM Treasury and Bank of England released the consultation for a Digital Pound by 2030.
Smart Money provides:
- Roaming bank accounts like roaming mobile phones
- Programmability to automate complex business workflows
- Businesses and consumers the ability to add transaction rules when you pay -eg split payments, automate VAT
- Safe - 100% backed by the central bank
Primary Motivations:
- Ensuring the role of the UK central bank is an anchor for confidence and safety in our monetary system
- Promoting innovation, choice and efficiency
Fostering stability and growth with access, innovation & choice.
- The digital GBP could improve resilience as an additional payment rail
- It could also complement existing financial inclusion initiatives
- Using the digital pound to improve cross-border payments is an opportunity but would take time and require int'l co-operation to deliver.
Now the technical bit of a CBDC.
Functional requirements needed to integrate with the existing financial system and payment rails and design considerations needed for a CBDC:
• Privacy
• Security
• Resilience
• Performance
• Extensibility
• Energy usage
One of the most important topics and concerns for CBDCs has been Privacy.
The law and the same approach will still apply to CBDCs.
It will be the same as today's AML/KYC regulation where commercial banks are responsible for compliance and managing the relationship with their customers. Complying with the same rules and regulations as today with non-CBDC money.
Smart Money provides:
- Roaming bank accounts like roaming mobile phones
- Programmability to automate complex business workflows
- Businesses and consumers the ability to add transaction rules when you pay -eg split payments, automate VAT
- Safe - 100% backed by the central bank
Primary Motivations:
- Ensuring the role of the UK central bank is an anchor for confidence and safety in our monetary system
- Promoting innovation, choice and efficiency
Fostering stability and growth with access, innovation & choice.
- The digital GBP could improve resilience as an additional payment rail
- It could also complement existing financial inclusion initiatives
- Using the digital pound to improve cross-border payments is an opportunity but would take time and require int'l co-operation to deliver.
Now the technical bit of a CBDC.
Functional requirements needed to integrate with the existing financial system and payment rails and design considerations needed for a CBDC:
• Privacy
• Security
• Resilience
• Performance
• Extensibility
• Energy usage
One of the most important topics and concerns for CBDCs has been Privacy.
The law and the same approach will still apply to CBDCs.
It will be the same as today's AML/KYC regulation where commercial banks are responsible for compliance and managing the relationship with their customers. Complying with the same rules and regulations as today with non-CBDC money.
Telegram
All about AI, Web 3.0, BCI
Google Research dropped the code and paper for MetNet-2, a neural net that can predict rain up to 12 hours in advance. This beats current weather forecast models which are instead huge physics simulations run on supercomputers.
Nature
Deep learning for twelve hour precipitation forecasts
Nature Communications - Can AI learn from atmospheric data and improve weather forecasting? The neural network MetNet-2 achieves this by forecasting the fast changing variable of precipitation up...
👍2