where might be the archils heal. sudoku has less possible solutions than an equivalent Latin square. you can say half the numbers are 1's and the other half the numbers are 2 and make a sudoku grid of 1's and 2's, the question is how many of the numbers have to be one in order to trip the grid up and how many variations do you have to try to trip the main sudoku grid up. if you can do it with a few combinations of 2 1's and the rest 2's then sudoku can be solved in p time, but can you use 3 grids 2 1's to solve 3 1's which it's big O is in exponential time. if you can trip a 3 1's system up with 2 1's systems then maybe you can trip up a sudoku with grids with 3 1's.
@dr3w97917 күн бұрын
Mr Tim Roughgarden, thank you for giving all of us the opportunity to learn for free what would normally cost thousands of dollars.
@harshitchauhan6334Ай бұрын
It's good to consider random leaders as balanced but what if we have 49% of byzantine nodes then there will be high probability of getting w unbalanced nodes
@antoine2571Ай бұрын
proof of the key lemma starts at 1:01:45
@antoine2571Ай бұрын
the audio is so good ! Thanks :) Btw the lecture is fantastic
@antoine2571Ай бұрын
Starts at 14:23 By the way this teacher is so good. I've never had such a good CS teacher unfortunately
@djeanqlАй бұрын
I cannot thank you enough for these quality videos, I appreciate you putting these on youtube!
@KhaledAlekasirАй бұрын
Thanks sir for all these great lectures you provide! always enjoying the way you teach.
@erigara3593Ай бұрын
Is my intuition into why initial configuration is ambiguous correct? Let’s assume that we have some protocol Pi and that under this protocol there is no ambiguous configurations. Since configuration is not ambiguous not future messages can influence the individual node decision. So it must make decisions only based on information it had initially (it’s private input). There is 4 ways we can transform private input into final decision: always 1, always 0, identity, inversion. And for each way transformation we can came up with such configuration where either validity or agreement is broken. Like for always 1 or 0 case validity is broken on all 0s or 1s respectively. For inversion validity and agreement is broken. And for identity agreement is broken when at least one node has different private input from other nodes. So we came to contradiction and initial configuration is ambiguous.
@TimRoughgardenАй бұрын
No, it's not that simple. Specifically a node with input 0 (say) may output 0 in some initial unambiguous configurations and 1 in other initial unambiguous configurations.
@erigara3593Ай бұрын
@@TimRoughgarden i see, so unambiguous configuration doesn’t mean that decision is already made by this node, it means that node eventually will terminate with certain value (0 or 1 depending on 0-configuration or 1-configuration). Thanks for great lectures!
@hahneortiz2 ай бұрын
@34:30 for case 2, is it probability alpha from state 1 to state 2, not to state 0?
@TimRoughgarden2 ай бұрын
Right, the slide is correct.
@hahneortiz2 ай бұрын
@10:15 I think it should be 20 or more based on 1000.
@hahneortiz2 ай бұрын
Why are we doubling the k in the theorem?
@cc98-oe7ol2 ай бұрын
Very intriguing lecture!
@DanSmith-ts5vs2 ай бұрын
Каково твое мнение о Покеткойне? Слышал, что это новая криптовалюта с инновационной технологией Proof-of-Stake. Стоит ли рассмотреть ее ближе?
@forheuristiclifeksh78363 ай бұрын
1:00
@amanarya50173 ай бұрын
By the end of this lecture I was like the professor deserves a Godel Prize for sure and guess what he is the recipient of the 2012 Godel Prize
@cc98-oe7ol2 ай бұрын
And he won it due to works on algorithmic game theory.
@cecilechau79323 ай бұрын
Single out message to ensure eventual delivery
@cecilechau79323 ай бұрын
FLP is the most importan result in distributed computing
@cecilechau79323 ай бұрын
Assumptions: no messages delays & no ordering - the protocol is event driven - weakest assumptions
@cecilechau79323 ай бұрын
Pki (key pairs + global share of public keys)
@cecilechau79323 ай бұрын
For BB problem or SMR f>1, you need >1 round of cross checking - for any additional one malicious note, one more rounding cross checking
@cecilechau79323 ай бұрын
Mostly validity and agreement
@viharivemuri72023 ай бұрын
It takes LogK bits to represent K, if the greedy algorithm is O(K) , isn't that itself an exponential time algorithm? How is this a polynomial algorithm, but an N*C algorithm for knapsack makes it exponential?
@timroughgardenlectures18613 ай бұрын
The input includes \Omega(k) different numbers, which require \Omega(k) bits to represent
@cecilechau79323 ай бұрын
Honest is not about intention it’s about bahaviour
@shymaaarafat13423 ай бұрын
I think it would be a good research, may lead to interesting results, how this 5/9 limit and slashing applicability can be mapped to DAG consensus protocols, and also whether the 2-stage time saving versions (produced by Chainlink) can handle such idea? I mean if this haven't been pursued before, and is not trivial, but I'm missing it.
@esadecimale3 ай бұрын
Fascinating lectures and results, thanks for sharing
@RosaLei3 ай бұрын
You always expand my mind in interesting ways because I have to research and dissect a lot of what you present and say! I appreciate your work, and I still think it is strangely underrated. Meanings for things change, but it's comforting to hear your voice and see you in the familiar setting of your room there. Thank you, @TimRoughgardenLectures1861 😊 Professor Tim Roughgarden! 🎉
@ferociousjuggler26683 ай бұрын
Great stuff as always!
@DamnBoiya3 ай бұрын
Blockchain is the future!
@abdullahjeffers90984 ай бұрын
that definition is confusing me . ANY ANOLGIES to simplify what it means?
@AriB14234 ай бұрын
Where are we using the fact that **all** honest nodes must have seen the transaction? Is it the fact that we just want any successive honest leaders (we can do this because everyone has seen tx?). If it was only one leader that had seen the transaction, we're not guaranteed to get two honest nodes in a row if we select leaders in a round robin way that have both seen the transaction.
@timroughgardenlectures18614 ай бұрын
Right, it could be that the only honest node that knows about the transaction never gets to propose their own block (because they inherit the block proposal of a previous Byzantine leader). Whereas if the second honest node in a row knows about the transaction, they will get to propose their own block, which will include that transaction.
@someRandomHuman54894 ай бұрын
Can you share the lecture notes!
@lornestalker66794 ай бұрын
Just wanted to say thanks for all of your content on KZbin, Tim! I'm currently studying my Masters at UT and your content has been a huge help. Hope you're doing well and would love to see some more content at some point!
@rodrigoherrera73924 ай бұрын
Legend
@aribiswas68634 ай бұрын
It is not clear to me who maintains this list of (pk_i, q_i)_{i \in [n]} of active validators from which construct the probability distribution which tells us who the leader for the round might be? Don't you need another consensus protocol for everyone to agree what the status of this list is? If we assume that is owned by a trusted third party, is this really decentralised?
@luisrnandezc5 ай бұрын
I'm just finishing the last week of the final course from the Coursera Algorithms Specialization. You are a gifted instructor and I'm grateful for the effort into making this knowledge available. Algorithm analysis and design is indeed a fascinating subject.
@timroughgardenlectures18614 ай бұрын
Thanks!
@swapnilaggarwal5 ай бұрын
thank you sir , love from India
@forheuristiclifeksh78365 ай бұрын
1:00
@doneilchapman15425 ай бұрын
Great video!
@imajinl.5 ай бұрын
Thanks for this lecture, Tim.
@lucaswhite17165 ай бұрын
why here are there no comments? this course is very well~
@cc98-oe7ol3 ай бұрын
This series is available on Coursera: Algorithm Specialization. I think most people just learn on Coursera.
@richdobbs65956 ай бұрын
Dang, I came here thinking I was going to hear a modern, high tech song equivalent to Riders in the Storm. Sigh.
@OsegueraJA6 ай бұрын
Awesome video! Thanks!
@chillydoog6 ай бұрын
1. Decentralized Digital Identity and Passport System (DDIPS) Concept: A blockchain-based platform that provides a sovereign digital identity to every individual. Unlike traditional IDs, these digital identities are user-owned, cannot be forged, and are recognized internationally. This system can integrate with government services, healthcare, finance, and cross-border mobility, enabling seamless, secure, and efficient verification processes. Unique Proposition: Utilizes the blockchain's no-owner characteristic to ensure neutrality and global accessibility, supporting user-owned data and enforcing property rights over personal information. 2. Global Real Estate and Asset Tokenization Platform (GRATP) Concept: A platform that democratizes access to real estate investments worldwide through tokenization. Users can buy and sell fractions of properties or physical assets as easily as purchasing stocks, completely on a blockchain infrastructure. This idea leverages smart contracts for trustless transactions and transparent ownership records. Unique Proposition: Opens up global investment opportunities to the average person with minimal capital, breaking down barriers imposed by traditional real estate investment structures. 3. Decentralized Autonomous Legal Entities (DALE) Concept: A framework for creating and managing legal entities entirely on the blockchain. These entities can enter contracts, own assets, and even sue or be sued, all governed by code. This concept challenges the traditional notion of corporations and legal personhood, offering a new way to think about organizational structures and responsibilities. Unique Proposition: Reduces operational friction and legal costs, providing a new method for international business operations without the need for complex legal compliance and intermediaries. 4. Blockchain-Based Universal Basic Income (UBI) System Concept: Utilizing blockchain to distribute a Universal Basic Income efficiently and transparently to citizens around the globe. This system could be funded through various mechanisms, including transaction fees, philanthropic contributions, or even a decentralized autonomous organization (DAO) that invests in other blockchain projects and distributes profits as UBI. Unique Proposition: Provides a scalable, tamper-proof method to address global inequality and poverty, showcasing the blockchain's potential for social good. 5. Decentralized Supply Chain and Authenticity Verification (DSCAV) Concept: A comprehensive platform that leverages blockchain for end-to-end visibility in supply chains, enabling consumers to verify the authenticity, origin, and ethical standards of products by scanning a QR code. This can apply to luxury goods, pharmaceuticals, food, and more, ensuring transparency and trust. Unique Proposition: Combats counterfeiting, ensures ethical sourcing, and builds consumer trust in brands by providing an immutable record of a product's journey from production to sale. 6. Energy Trading and Carbon Credit Marketplace (ETCCM) Concept: A decentralized platform that facilitates the trading of renewable energy between producers and consumers directly and issues verifiable carbon credits for reductions in greenhouse gas emissions. This can incentivize sustainable practices among businesses and individuals alike. Unique Proposition: Encourages renewable energy production and consumption, contributing to the fight against climate change by making carbon credit trading accessible, transparent, and efficient.
@ahmedfakhrudeen6 ай бұрын
Hello, thanks for the valuable videos. Can you please share with the text of the videos. Regards. Ahmed
@ursamajor32756 ай бұрын
what is the definition of an "Ambiguous Configuration"?
@42770991228 күн бұрын
It's defined in the previous video kzbin.info/www/bejne/rpLQaJubatKooa8si=3oQFxV78iqipQV6L
@ursamajor32756 ай бұрын
for step 2 @ 20 min 20 secs, Can you explain further how it it does not increase the chance of winning and is equivalent to someone with only 1 chance?
@TimRoughgarden6 ай бұрын
e.g. if sampling probability for one coin is 10% and you have two coins, you'll get sampled with probability Pr[at least one of two independent coins would be sampled] = 2/10 - (1/10)^2 = 19%
@ursamajor32756 ай бұрын
@@TimRoughgarden from what you said, 10% has increased to 19%. Hence, there is an increased chance of winning. Hence, how is there a claim that there isn't?
@timroughgardenlectures18616 ай бұрын
@@ursamajor3275 The point is that staking two coins under one identity has the same effect as creating two identities and staking one coin under each.
@ursamajor32756 ай бұрын
@@timroughgardenlectures1861 the probability shown increased. The 2nd coin did increase the probability from 10% to 19% as you had shown. Hence, I am confused. How is this the same? Is there a clearer example where mathematically they are equivalent? that is P(2 coins under 1 identity) = P(creating 2 identities and staking 1 coin under each)?
@billwu73677 ай бұрын
you are the king!
@hc20797 ай бұрын
I want to thank you very (very) much!! i'm studying all your videos one after the other; i'm studying blockchain and your courses are more than helpful!!
@ZackFan7 ай бұрын
Gizmo
@matteogirelli10237 ай бұрын
I feel like in this lecture there must be the largest amount of as obnoxious as irrelevant questions asked to a teacher I have ever seen. Just plain pointless. They can't even see that the match is an example to explain the algorithmic nature of strategic decision making. He is a professor, not a badminton referee for God's sake!