La vigne au 22 mai 2021

 

Avant le passage de la tondeuse

Après le passage de la tondeuse. Reste à faire entre les ceps.

J'ai prélevé vingt feuilles pour faire une analyse par fluorimétrie

Les résultats de l'analyse : il y a une carence en azote

La vigne au 14 mai 2021

 Quelques photos de la vigne au 14 mai :

.

On voit les raisins en formation

J’ai profité de l’éclaircie pour « monder » la parcelle

Vu de la colline de Brouilly depuis la parcelle

J’ai compté 9 raisins sur un cep

Polygon

• Polygon is a Commit Chain.
• Commit chains (extract from Mastering Blockchain):
• are a more generic term for Vitalik Buterin's plasma proposal. (http://plasma.io)
• are referred as non-custodial side chains but without a new consensus mechanism, as it is the case for side chains. They rely on main chain consensus mechanism, so they can considered as safe as the main chain. The operator of the commit chain is responsible for facilitating communication between transacting participants and sending regular updates to the main chain. See https://eprint.iacr.org/2018/642.pdf

Useful links

• https://ethhole.link/
• https://twitter.com/0xPolygon

Mastering Blockchain - Imran Bashir - Part II

 

 

Chapter 11: Ethereum 101

• Vitalik Buterin conceptualized Ethereum (https://ethereum.org)
• Decentralized Applications (DApps): this concept is in contrats to Bitcoin, where the scripting language is limited and only allows necessary operations.
• Ethereum 1.x: Proof-of-Work (PoW)
• Ethereum 2.x: Proof-of-Stake (PoS)
• Serenity release is ETH2.0.
• Ethereum, just like any blockchain, can be visualized as a transaction-based state machine.
• The transaction is digitally signed by the sender as proof that he is the owner of the ether.
• The transaction is then picked up by nodes called miners on the Ethereum network for verification and inclusion in the block.
• The details of the transaction can be visualized: https://etherscan.io/
• This hash is the unique ID of the transaction that can be used to trace this transaction throughout the blockchain network.
• The Ethereum network is a peer-to-peer network where nodes participates in order to maintain the blockchain and contribute to the consensus algorithm.
• Ethereum client is Geth: provides various functions such as mining and account management.
• Keys and addresses: The private key is a generated randomly and is kept secret; whereas public key is derived from the private key. Addresses are derived from public keys and are 20-byte codes to identify accounts. 
• Accounts are one of the main building blocks of the Ethereum blockchain. They are defined by pairs of private and public keys.
• Ethereum is a transaction-driven state machine.
• In case of transaction failure due to insufficient account balance or gas, all state changes are rolled back except for the feee payment, which is paid to the miners.
• The Contracts Accounts: the code is executed by the EVM (Ethereum Virtual Machine) by each mining node on the Ethereum network.
• Accounts allow interaction with the blockchain via transactions.
• A transaction in Ethereum is a digitally signed data packet using a private key that contains the instructions that either result in a message call or contract creation.
• Nonce: the nonce is a number that is incremented by one every time a transaction is sent by the sender. It must be equal to the number of transactions sent and is used as a unique identifier for the transactions.
• Gas price: the gas price represents the amount of Wei required to execute the transaction.
• A block is a data structure that contains batches of transactions. Transactions can be found in either transactions pools or blocks. In transaction pools, they wait for verification by a node, and in blocks, they are added after successful verification.
• Ehtereum's PoW is called Ethash and was originally designed to be ASIC resistant.
• Recursive Length Prefix (RLP) is linked to serialization and deserialization which are also referred as marshaling and un-marshaling respectively. Some commonly used serialization formats include XML, JSON, YAML, protocol buffers and XDR.
• Message calls result in a state transition. A message call is the act of passing a message from one account to another. If the destination account as an associated EVM code, then the EVM will start upon the receipt of the message to perform the required operations.
• Messages are similar to transactions; however, the main difference is that they are produced by the contracts.
• A call basically runs message call transactions locally on the client and never costs gas nor results in a state change.
• At a fundamental level, the Ethereum blockchain is a transaction and consensus-driven state machine.
• Code hash: this is an immutable field that contains the hash of the smart contract code that is associated with the account.
• After the Decentralized Autonomous Organization (DAO) hack, a hard fork was proposed: Ethereum classic and its currency represented by ETC, whereas the hard-forked version is ETH.
• The EVM is a simple stack-based execution machine that runs byte code instructions to transferm the system state from one state to another. Last In First Out (LIFO).
• EVM optimization is an active area of research.
• WebAssembly (Wasm) was developed by Google, Mozilla and Microsoft.
• Precompiled contracts can potentially become native extensions and might be included in the EVM opcodes in the future.
• Next hard fork release of Ethereum called Berlin will introduce new pre-compiled contracts for BLS12-381 curve operations with other updates.

Links

• https://ethereum.org
• https://etherscan.io/

Acronyms

• OOG: Out of Gas

Translation

• immutable: immuable
• 
  

Mastering Blockchain - Imran Bashir

 

• Préambule

• Un livre référence pour qui s'intéresse à la blockchain d'un point de vue fondamental.
• Un mois de lecture du 12 mars au 14 avril.
• Une approche approfondie qui va des bases du concept de livre des transactions partagées, aux protocoles de consensus, aux applications distribuées, aux cryptos monnaies, à l'ethereum, au web3 etc...
• Mes notes de lecture en résumé ci-dessous.

• Chapter 1: Blockchain 101

• Multiple confirmations of a transaction over time provide consistency in Bitcoin. For this purpose, the process of mining was introduced in Bitcoin. 
• Mining is a process that facilitates the achievement of consensus by using the PoW algorithm.
• With physical cash, it is almost impossible to trace back spending to the individual who actually paid the money.
• Smart contract: code needed to execute a required function when certain conditions are met.
• Smart property: it is possible to link a digital or physical asset to the blockchain in such a secure and precise manner that it cannot be claimed by anyone else.
• Enterprise Ethereum Alliance
• Hyperledger

• Chapter 2: Decentralization
• Many exiting applications and ideas emerge from the decentralized blockchain technology, such as decentralized finance and decentralized identity.
• A significant innovation in the decentralized paradigm is decentralized consensus.
• Blockchain can serve as a decentralized health record management system where health records can be exchanged securely  and directly between different entities.
• Not everything can or needs to be decentralized.
• There are systems that pre-date blockchain and Bitcoin, including BitTorrent and the Gnutella file-sharing system.
• IPFS and BigChainDB are more suitable for storing large amounts of data in a decentralized way.
• Many services hat a government commonly offers can be delivered via blockchains such as government identity card systems, passports, and records of deeds, marriages and births.
• Web 3 = decentralized internet

• Chapter 3: Symmetric cryptography

• Symmetric cryptography refers to a type of cryptography where the key that is used to encrypt the data is the same one that is used for decrypting the data.
• Hash functions play a vital role in blockchain.
• The SHA-256 algorithm is used in Bitcoin's PoW algorithm.
• Chapter 4: Public key cryptography
• Asymmetric cryptography refers to a type of cryptography where the key that is used to encrypt the data is different from the key that is used to decrypt the data. These keys are called private and public keys.
• Digital signatures are used to provide data origin authentication and non-repudiation.
• Hash functions are used to build Merkle trees.
• Hash functions are of particular importance in blockchains, as they are key to constructing Merkle trees, which are used in blockchains for the efficient and fast verification of large datasets.

• Chapter 5: Consensus algorithm
• In distributed systems, a common goal is to achieve consensus (agreement) among nodes on the network even in the presence of faults.
• State machine replication is a standard technique to achieve fault tolerance in distributed systems.
• In an asynchronous environment, the deterministic consensus is impossible.
• Liveness means that the protocol can make progress even if the network conditions are not ideal.
• PBFT (Practical Byzantine Fault Tolerant) provides immediate and deterministic transaction finality. This is in contrast with the PoW protocol, where a number of confirmations are required to finalize a transaction with high probability.
• IBFT (Istanbul Byzantine Fault Tolerant): validators can be added or removed by voting between members of the network.
• Nakamoto consensus, or PoW, is the longest-running blockchain network. 
• The PoW is designed to mitigate Sybil attacks. 
• A Sybil attack is a type of attack that aims to gain a majority influence on the network to control the network. Its a type of attack where a single adversary creates a large number of nodes with fakes identities on the network, which are used to gain influence on the network. This attack is also prevented in Bitcoin by using PoW, where miners are required to consume a considerable amount of computing power to earn rewards.
• The key idea behind PoW as a solution to the Byzantine generals problem is that all honest generals (miners in the Bitcoin world) achieve agreement on the same state (decision value). Note that this is a probabilistic solution and not deterministic.
• PoS (Proof of Stake) mechanisms generally select a stake holder and grant appropriate rights based on their staked assets.
• Create a system that meets all the requirements without compromising the core safety and liveness properties of the system.

• Chapter 6: Introducing Bitcoin

• Bitcoin is quite unstable and highly volatile.
• PoW concept: the money is created by broadcasting the solution to a previously unsolved computational problem. PoW is used to secure the blockchain. This is a proof that enough  computational resources have been spent in order to build a valid block.
• The PoW problem, also known as the partial hash inversion problem consumes a high amount of resources, including computing power and electricity. This process also secures the system against fraud and double-spending attacks while adding more virtual currency to the Bitcoin ecosystem.
• Hal Finney was the first person to receive Bitcoin from Satoshi Nakamoto.
• In the Bitcoin network, users have to wait for six blocks, which is equivalent to an hour, to get the right level of confidence that a transaction is final.
• Trustless refers to the distribution of trust between users, rather than a central entity.
• Network is owned collectively by its users instead of a single entity. The more users use the network, the more valuable it becomes.
• The double spending problem arises when a user sends coins to two different users at the same time and they are verified independently as valid transactions. The double spending problem is resolved in Bitcoin by using a distributed ledger (the blockchain) where every transaction is recorded permanently.
• The sender enters the receiver's address.
• The transaction fee is calculated based on the size of the transaction and on the fee rate.
• The higher the transaction fee, the greater the chance that your transaction will be picked up as a priority and included in the block.
• Bitcoin transactions are serialized for transmission over the network and encoded in hex format.
• Public keys can be represented in an uncompressed or compressed format, and are fundamentally x and y coordinates on an elliptic curve.
• Transaction malleability has been solved with the so-called SegWit soft-fork upgrade of the Bitcoin protocol.
• The paper wallets can be stored physically as an alternative to the electronic storage of private keys.
• Transactions are not encrypted and are publicly visible on the blockchain.
• The mining is the process through which the blockchain is secured and new coins are generated as a reward for the miners who spend appropriate computational resources. 
• Mining is a process by which new blocks are added to the blockchain. Blocks, once mined and verified, are added to the blockchain, which keeps the blockchain growing.
• Transaction pool, also known as memory pools, are basically created in local memory (computer RAM) by nodes (Bitcoin clients) in order to maintain a temporary list of transactions that have not yet been added to a block.
• The time for transaction confirmation usually ranges from 10 minutes to over 12 hours in some cases.
• Bitcoin uses a simple stack-based language called Script. Scripting is quite limited and can only be used to program one thing: the transfer of bitcoins from one address to other addresses.
• In simple words, locking means providing Bitcoins to somebody, whereas unlocking means consuming the acquired Bitcoins.
• A coinbase transaction or generation transaction is always created by a miner and is the first transaction in a block. It is used to create new coins.
• Nonce is an arbitrary number that miners change repeatedly to produce a hash that is lower than the difficulty target.
• New blocks are added to the blockchain approximately every 10 minutes.
• Miners are rewarded with new coins if and when they discover new blocks by solving the PoW. 
• The rate of creation of new bitcoins decreases by 50% every 210000 blocks, which is roughly every four years.
• After 210000 blocks, the block reward halves. In November 2021 it halved down to 25 bitcoins. Currently, since May 2020, it is 6.25 bitcoins per block.
• Mining difficulty increases over time and bitcoins that could once be mined by a single-cpu laptop computer now require dedicated mining centers to solve the hash puzzle.
• The difficulty is calculated every 2016 blocks (around two weeks) and adjusted accordingly.
• The Bitcoins miners have to calculate hashes to solve the PoW algorithm.
• The hash rate basically represents the rate of hash calculation per second.
• As the core principle behind mining is based on the double SHA256 algorithm, over time, experts have developed sophisticated systems to calculate the hash faster and faster. 
• ASICs were designed to perform SHA-256 operations.
• Currently mining is out of the reach of individuals due to the vast amount of energy and money needed to be spent in order to build a profitable mining platform.
• Solo mining is not very profitable now.
• Core
• https://btc.com

• Chapter 7: the Bitcoin network and payment

• The Bitcoin network is a peer-to-peer (P2P) network where nodes perform transactions.
• There are different types of nodes on the network. The two main types of nodes are full nodes and simple payment verification (SPV) nodes.
• SPV nodes only download the headers of the blocks while syncing with the network.
• SPV clients are used to verify payments without requiring the download of the full blockchain.
• On a Bitcoin network, there are full clients (nodes), which perform the function of storing a complete blockchain. If you cannot run a full node, then SPV clients can be used to verify that particular transactions are present in the block by only downloading the block headers instead of the entire blockchain. At times, even running an SPV node is not feasible (especially on low-resource devices such as mobile phone) and the requirement is only to send and receive Bitcoin somehow. For tis purpose, wallets (wallet software) are used that do not require downloading even the block headers.
• Since version 0.10.0, the initial block download method named headers-first was introduced. This resulted in major performance improvement, and blockchain synchronization that used to take days to complete started taking only a few hours.
• The wallet software is used to generate and store cryptographic keys. It performs various useful functions, such as receiving and sending Bitcoin, backing up keys, and keeping track of the balance available. 
• Private keys are used by wallets to sign the outgoing transactions. Wallet do not store any coins, and there is no concept of wallets storing balance or coins for a user. In fact, in the Bitcoin network, coins do not exist; instead, only transaction information is stored on the blockchain (more precisely, UTXO, unspent outputs), which are then used to calculate the number of bitcoins.
• Paper wallets can be generated on line from various service providers such as https://bitcoinpaperwallet.com or https://www.bitaddress.org/
• Lightning network is a solution for scalable off-blockchain instant payments. It was introduced in early 2016 and allows off-blockchain payments.
• Innovation in Bitcoin: improvement proposals are usually made in the form of Bitcoin Improvement proposals (BIPs).
• Excellent ideas have emerged from research and innovation efforts related to Bitcoin.
• The Bitcoin network can only process approximately three to seven transactions per second which is a tiny number compared to Visa network which can process on average 24,000 transactions per second. PayPal can process approximately 200 transactions per second whereas Ethereum can process up to, on average 20.
• The Segregated Witness (SegWit): the key idea behind SegWit is the separation of signature data from transaction data (that is, a transaction Merkle tree), which results in the size of the transaction being reduced.
• Not only has the original Bitcoin evolved quite significantly since its introduction, but there are also new blockchains that are either forks of Bitcoin or novel implementations of the Bitcoin protocol with advanced features.
• Bitcoin cash (BCH) uses Proof of Work (PoW) as a consensus algorithm, and mining hardware is still ASIC-based.
• Bitcoin Gold is hard fork and uses the Equihash algorithm as its mining algorithm instead of PoW.
• Market orders means that as soon as the prices match, the order will be fulfilled immediately. Limit orders allow for buying and selling a set number of bitcoins at a specified price or better.

• Chapter 8: Bitcoin Clients and APIs

• The Bitcoin client can be installed from https://bitcoin.org/en/download
• A Bitcoin client is a piece of software that is responsible for generating private/public key pairs and facilitates Bitcoin payments using the Bitcoin blockchain.
• The Bitcoin source code can be downloaded and compiled if users wish to participate in the Bitcoin code or for learning purposes.
• Regtest mode (regression testing mode) can be used to create a local private blockchain for testing purposes.
• Using the HTTP REST interface available at https://bitcoincore.org/en/doc/0.19.0/rpc
• Blockchain info: https://blockchain.info/api
• BitPay: https://bitpay.com/api
• Block.io: https://www.block.io

• Chapter 9: Alternative Coins
• Alternative approaches to Bitcoin can be divided broadly into two categories, based on the primary purpose of their development. If the primary goal is to build a decentralized blockchain platform, they are called alternative chains; if the sole purpose of the alternative project is to introduce a new virtual currency, it is called an altcoin.
• Namecoin: the primary purpose is to provide decentralized naming and identity services instead.
• Ethereum classic is the old chain and Ethereum is the new chain after the fork. Such a contentious hard fork is not desirable.
• Altcoins must be able to attract new users, trades and miners.
• Currency gains its value, especially in the virtual currency space, due to the network effect and its acceptability by the community.
• The power is shifting toward miners or mining pools who can afford large-scale ASIC farms. This power shift challenges the core philosophy of the decentralization of Bitcoin.
• PoW does have various drawbacks, and the biggest of all is energy consumption. It is estimated that the total electricity consumed by Bitcoin miners currently is more than that of Greece at 59.61 Terawatt hash (TWh)
• Proof of Stake (PoS): users are required to demonstrate the possession of a certain amount of currency (coins), thus proving that they have a stake in the coin.
• In Bitcoin, the difficulty is adjusted every 2016 blocks.
• Privacy and anonymity: as the blockchain is a public ledger of all transactions and is openly available, it becomes easy to analyze it. This is a big concern from a privacy point of view. Various proposals have been made to address the privacy issue in Bitcoin. These proposals fall into three categories: mixing protocols, third-party mixing networks and inherent anonymity.
• Dandelion is a proposal that aims to make transactions on a Bitcoin network untraceable. The Dandelion protocol is composed of an anonymity phase and a spreading.
• Colored coins: coloring a bitcoin refers colloquially to updating it with some metadata representing a digital asset (smart property). The coin still works and operates as a Bitcoin, but additionally carries some metadata that represents some assets.
• Colored coins can be used to represent a multitude of assets, including, but not limited to, commodities, certificates, shares, bonds and voting.
• Counterparty is another service that can be used to create custom tokens.
• Counterparty works based on the same idea as colored coins by embedding data into regular Bitcoins transactions, but provides a much more productive library and a set of powerful tools to support the handling of digital assets.
• Counterparty allows the development of smart contracts on Ethereum using the Solidity language and allows interaction with the Bitcoin blockchain.
• Development of altcoins: Altcoin projects can be started very quickly from a coding point of view by simply forking the Bitcoin or another coin's source code, but this is probably not enough. Usually, the code base is written in C++, as was the case for Bitcoin, but almost any language can be used to develop coin projects; for example Golang or Rust.
• Reward halving rate: Bitcoin is halved every 4 years and now is set to 12.5 Bitcoin.
• Total supply of coins: this number sets the total limit of the coins that can be ever be generated.In Bitcoin, the limit is 21 million, whereas in Dogecoin, it's unlimited.
• Token versus cryptocurrency: A cryptocurrency is a native coin for a standalone blockchain. In contrast, a token is a representation of the value of some asset.
• Initial Coin Offering (ICO) is a method for crowdfunding. The first successful ICO was that of Ethereum, which raised 18 million USD in 2014. A recent success is Tezos, which made 232 million USD in a few weeks. Another example is Filecoin, which raised more than 250 million USD. A more recent example is EOS, which raised a record of over 4 billion USD.
• Ethereum has become a platform of choice for ICOs due to its ability to create new tokens and with the ERC20 (Ethereum Request for Comments)standard, it has become even more accessible.
• Cryptocurrencies are a very attractive area for research.
• Chapter 10: Smart Contracts

• Smart contracts are now on going and intense area of research in the blockchain space.
• Smart contracts can bring many benefits such ad increased security, cost-saving and transparency to industries (especially the finance industry).
• Nick Szabo forts theorized smart contracts in the 1990's.
• A smart contract is a secure and unstoppable computer program representing an agreement that is automatically executable and enforceable.
• Code is the law.
• Certain inputs that need to be provided by people can and should be also automated. Oracles can be used for that purpose.
• English law recognized crypto assets as traceable property and smart contracts as enforceable agreements. This announcement was made by the UK Jurisdiction Taskforce (UKJT) (https://technation.io/news/uk-takes-significant-step-in-legal-certainty-for-smart-contracts-and-cryptocurrencies)
• If humans and machines can both understand the code written in a smart contract, it might become acceptable in legal situations, as opposed to just a piece of code that no one understands except for programmers.
• Legal Knowledge Interchange Format (LKIF)
• Smart contract have the following properties:
• automatically executable
• enforceable
• deterministic
• semantically sound
• unstop-able
• Ricardian contracts (https://iang.org/papers/fc7.html) fundamental: write a document that is understood and accepted by both a court of law and computer software.
• Consider a smart contract as a standalone entity that is capable of encoding legal prose and code (business logic)
• Solidity is a language that has been introduced with the Ethereum blockchain to write smart contracts.
• Tibco StreamBase product is a Java based system used for building event-driven, high frequency trading systems is an equivalent to DApp (Decentralized App)
• An inherent limitation with smart contracts is that they are unable to access any external data. The concept of oracles was introduced to address this issue. An oracle is an off-chain source of information that provides the required information to the smart contracts on the blockchain.
• The limitation with smart contracts is that they cannot access external data because blockchains are closed systems without any access to the real world.
• An oracle can be defined as an interface that delivers data from an external source to smart contracts.
• TLS notary provides a piece of irrefutable evidence to an auditor that specific web traffic has occurred between a client and a server. In is based on TLS (Transport Layer Security).
• To prove the authenticity of the data retrieved by oracles from external resources, attestation such as the TLSNotary are used.
• Ledger proof, Ledger Nanao S: the primary purpose of these devices is a secure hardware cryptocurrency. These devices run a particular OS called Blockchain Open Ledger Operating system (BOLOS)
• A trusted hardware assisted proof is Town Crier (https://www.town-crier.org) which provides an authenticated data feed for smart contracts. It uses Intel SGX to provide a security guarantee that the requested data has come from an existing trustworthy source.
• Various other blockchain platforms support smart contracts such as Monax, Lisk, Counterparty, Stellar, Hyperledger fabric, Axoni core, Neo, EOSIO and Tezos.
• DAO: Decentralized Autonomous Organization was a smart contract written to provide a platform for investment.
• DAO: due to a bug, called the reentrancy bug, it was hacked in June 2016. An equivalent to approximately 3.6 million ether was siphoned out of the DAO into another account. This incident resulted in a hard fork on the Ethereum blockchain, which was introduced to recover from the attack. Subsequently, resistance against this hard fork resulted in the creation oof Ethereum Classic, where a large number of users decided to keep the mining on the old chain.
• The security of smart contracts is an area of deep interest for researchers.
• Ricardian contracts are concerned with the definition of the contract, whereas smart contracts are geared toward the actual execution of the contract.

Chapter 11: Ethereum 101

• Ethereum has its own decentralized and distributed ecosystem that uses Swarm for storage and the Whisper protocol for communication.
• USDT token (Tethers) using the ERC 20 standard.
• Merkle-Patricia trees are used in the Ethereum blockchain.
• Ethereum with its Serenity release will soon transition to a PoS based consensus algorithm.

• DeFi

• https://defipulse.com

• Des liens nouveaux et intéressants 

• https://steemit.com
• https://ipfs.io
• https://identity.foundation
• https://openbazaar.org
• https://www.wireshark.org

• Acronymes

• AML: Anti Money Laundering
• BFT: Byzantine Fault Tolerance
• CAP Theorem : Consistency, Availability and Partition
• DAO: Decentralized Autonomous Organization
• EVM : Ethereum Virtual Machine
• ECC: Elliptic Curve Cryptography
• KYC: Know Your Customer
• Nonce: a nonce is a number that is generated and used only once.
• Merkle root: In a blockchain block, it is the combined hash of the transactions in the block
• PBFT: Practical Byzantine Fault Tolerance

• Translation

• escrow: dépôt
• unredeemable: non réclamable
• tumbler: acrobate

Anti-Fragile Nassim Nicholas Taleb

 

• The Black Swan problem: the impossibility of calculating the risks of consequential rare events and predicting their occurence.
• Heuristics are simplified rules of thumb that make things simple and easy to implement.
• One's personal experience gives the stamp of authenticity and sincerity of opinion.
• Abundance is harder for us to handle than scarcity.
• Using error as a source of information.
• What Erasmus called ingratitudo vulgi, the ingratitude of the masses, is increasing in the age of the globalization and the Internet.
• Techme: crafts and know how, not episteme: book knowledge, know what.
• Principle of subsidiarity: things should be handled by the smallest possible unit that can manage them with efficacy.
• It seems, wore Machiavelli, that in the midst of murders and civil wars, our republic became stronger and its citizens infused with virtues...A little bit of agitation gives resources to souls and what makes the species prosper isn't peace, but freedom.
• One of life's packages: no stability without volatility
• And the final lesson is that one should not expect laurels for bringing the truth.
• People take more risks when supply with a forecast.
• Fail early (so they can start again)
• Fooled by randomness: we have a tendency to underestimate the role of randomness in human affairs.
• Avoid being blind to the natural antifragility of systems, their ability to take care of themselves, and fight our tendency to harm and fragile them by not giving a chance to do so.
• Accès to data increases intervention.
• Junk food is iatrogenic.
• Too much information becomes harmful.
• People had real links, unlike in social networks.
• There are ample empirical findings to the effect that providing someone with a random numerical forecast increases his risk taking, even if the person knows that the projections are random.
• If you have extra cash in the bank, you don't need to know with precision which event will cause potential difficulties.
• We should first make things more robust to defects and forecasts errors.
• The mother of all stressors called time.
• Books, being robust, are immune to power outages.
• Focus on exposure to failure.
• Warren Buffet tries to invest in businesses that are "so wonderful that an idiot can run them. Because sooner or later, one will".
• Excess wealth, if you don't need it, is a heavy burden.
• Curiosity is antifragile, like an addiction, and is magnified by attempts to satisfy it, books have a secret mission and ability to multiply, as everyone who has wall-to-wall bookshelves knows well.
• Stay robust to how others treat you.
• Someone who predicts will be fragile to prediction errors.
• Robustness: protection against harm from emotions.
• Success brings an asymmetry: you now have a lot more to lose than to gain. You are hence fragile.
• Men feel the good less intensely than the bad.
• When I was trader, a profession rife with a high dose of randomness, with continuous psychological harm that drills deep into one's soul.
• Wealth is the slave of the wise man and master of the fool.
• You are anti fragile for a source of volatility if potential gains exceed potential losses (and vice versa)
• Nothing can be done both hastily and safely, almost nothing.
• Provide for the worst; the best can take care of itself.
• When I left the office, my day job ceased to exist until the next day and I was completely free to pursue what I found most valuable and interesting.
• And professions can be serial: something very safe, then something speculative.
• Montaigne have done a serial barbell: pure action, then pure reflection.
• Or, if I have to work, I find it preferable (and less painful) to work intensely for very short hours, then do nothing for the rest of the time (assuming doing nothing, is really doing nothing).
• If you dislike someone, leave him alone or eliminate him; don't attack him verbally.
• Reduced pains from adverse events, while keeping the benefits of potential gains.
• So just as Stoicism is the domestication, not the elimination, of emotions, so is the barbell a domestication, not the elimination, of uncertainty.
• An agent does not move except out of intention to an end.
• So let us call the teleological fallacy the illusion that you know exactly where you are going, and that you know exactly where you were going in the past, and that others have succeeded in the past by knowing where they were going.
• Flâneur continuously, and what is crucial, rationally, modifies his targets as he acquires information.
• People don't know what they want until you provide them with it.
• The advantages of wealth: the most important one being independence and the ability to only occupy your mind with matters that interest you.
• Limited loss and large possible outcome.
• And if you make more when you are right than you are hurt when you are wrong, then you will benefit, in the long run, from volatility (and the reverse).
• You will never get to know yourself, your real preferences, unless you face options and choices.
• "long gamma" means benefits from volatility and variability.
• An option is the weapon of anti fragility.
• The option is a substitute for knowledge.
• People look for books that support their mental program.
• Necessity is the mother of invention.
• Poverty makes experience.
• Good speculative bets come to you, you don't get them just by staying focused on the news.
• A way to domesticate uncertainty is to work rationally without understanding the future.
• You cannot look at the future by naive projection of the past.
• This brings us to the difference between doing and thinking.
• In theory there is no difference between theory and practice, in practice there is.
• Practitioners don't write; they do. Birds fly and those who lecture them are the ones who write their story. So it is easy to see that history is truly written by losers with time on their hands and a protected academic position.
• Small amounts per trial, lots of trials, broader than you want. Why? Because in Extremistan, it is more important to be in something in a small amount than to miss it. As one venture capitalist told me "The payoff can be so large that you can't afford not to be in everything".
• Consider the difference between lions in the wild and lions in captivity. Lions in captivity live longer; they are technically richer, and they are guaranteed job security for life, if these are the criteria you are focusing on...
• When I worked in trading rooms, where you sit most of the time waiting for things to happen, a situation similar to that of people sitting in bars or mafia men "hanging around".
• Avoidance of boredom is the only worthy mode of action. like otherwise is not worth living.
• I ordered most of every book with "probability" or "stochastic" in its title. I read nothing else for a couple of years, no course material, no newspaper, no literature, no nothing. Five years later I was set for life.
• Much of what other people know isn't worth knowing.
• What I decided to read on my own, I still remember.
• But Fat Tony's power in life is that he never lets the other person frame the question. He taught Nero that an answer is planted in every question; never respond with a straight answer to a question that makes no sense to you.
• Traditions provide an aggregation of filtered collective knowledge.
• Friedrich Hayek: the pricing system reveals through transactions the knowledge embedded in the society.
• Replacing our currency holdings with precious metals, as there are not owned by governments.
• The need to focus on the payoff from your actions instead of studying the structure of the world.
• The payoff, what happens to you (the benefits or harm from it), is always the most important thing, not the event itself.
• Education is an institution that has been growing without external stressors; eventually the thing will collapse.
• When a collection of people write "there is nothing new here" and each one cites a different originator of the idea, one can safely say there is something effectively new.
• For the fragile, shocks bring higher harm as their intensity increases.
• If you earn more than you lose from fluctuations, you want a lot of fluctuations.
• Small may be beautiful, it is certainly less fragile.
• It is completely wrong to use the calculus of benefits without including the probability of failure.
• As with the European Union's subsidiary principle, "small" here means the smallest possible unit for a given function or task that can operate with a certain level of efficiency.
• The idea that smooth functioning at regular times is different from the rough functioning at times of stress.
• Planning fallacy.
• This has led me to install a governmental golden rule: no borrowing allowed, forced fiscal balance.
• Globalization has had the effect of making contagions planetary.
• The technique, detecting acceleration of harm, applies to anything that entails decision making under uncertainty, and risk management.
• The notion of average is of no significance when one is fragile to variations.
• The more uncertainty, the more for optionality to kick in, and the more you will outperform. This property is very central to life.
• Remember from the logic of the barbell that it is necessary to first remove fragilities.
• We know a lot more what is wrong than what is right.
• Negative knowledge (what is wrong, what does not work) is more robust to error than positive knowledge (what is right, what works).
• In general, failure (and disconfirmation) are more informative than success and confirmation, which is why I claim that negative knowledge is just more "robust".
• Paul Valéry : "Que de choses il faut ignorer pour agir".
• Obvious decisions (robust to error) require no more than a single reason.
• Antifragility implies, contrary too initial instinct, that the old is superior to the new, and much more than you think.
• Recall the foundation asymmetry: the anti fragile benefits from volatility and disorder, the fragile is harmed. Well, time is the same as disorder.
• The large multinational bureaucratic corporation with "empty suits" at the top.
• Much progress comes from the young because of their relative freedom from the system and courage to take action that older people lose as they become trapped in life.
• New technology is easier to hype up.
• The future is in the past. Actually there is an Arabic proverb to that effect: he who does not have a past has no future.
• Information has a nasty property: it hides failure. Many people have been drawn to, say, financial markets after hearing success stories of someone getting rich in the stock market and building a large mansion across the street, but since failures are buried and we don't hear about them, investors are led to overestimate their chances of success.
• If you announce to someone "you lost 10 000€", he will be much more upset than if you tell him "your portfolio value, which was 785 000€, is now 775 000€".
• Our brains have a predilection for shortcuts, and the variation is easier to notice (and store) than the entire record. It requires less memory storage. This psychological heuristic (often operating without awareness), the error of variation in place of total, is quite pervasive, even with matters that are visual.
• We notice what varies and changes more than what plays a large role but doesn't change. We rely more on water than on cell phones but because water does not change and cell phones do, we are prone to thinking that cell phones play a larger role than they do.
• Treadmill effects: we notice differences and become dissatisfied with some items and some classes of goods.
• If something that makes no sense to you (say, religion if you are an atheist, or some age-old habit or practice called irrational); if that something has been around for a very, very long time, then, irrational or not, you can expect it to stick around much longer, and outlive those who call for its demise.
• The link everyone is obsessing about between salt and blood pressure has no statistical basis.
• What I am against is naive rationalized, pseudo learned discourse, with green lumber problems, one that focuses solely on the known and ignores the unknown.
• Whenever possible, replace the doctor with human anti fragility. But otherwise don't be shy with aggressive treatments.
• I derived the rule that what is called "healthy" is generally unhealthy, just a "social" networks are antisocial, and the "knowledge" based economy is typically ignorant.
• In a large set of circumstances (marginal diseases), anything that takes you away from the doctor and allows you to do nothing (hence gives nature a chance to do its work) will be beneficial.
• When we are herbivores, we eat steadily; but when we are predators we eat more randomly.
• Human response to hunger: that biological mechanisms are activated by food deprivation.
• Religion with ritual fasts have more answers than assumed by those who look at them too literally.
• The economic system is loading future generations with public governmental debt, causing depletion of resources, and environment blight (fléau) to satisfy the requirements of the security analysts and the banking establishment (once again, we cannot separate fragility from ethics)
• Place aux autres.
• Ethics. Under opacity and in the newfound complexity of the world, people can hide risks and hurt others, with the law incapable of catching them. Iatrogenics has both delayed and invisible consequences. It is hard to see causal links, to fully understand what's going on.
• If we are today, it is because someone, at some stage, took some risks for us.
• There is a pseudo courage that comes from risk blindness, in which people underestimate the odds of failure. We have ample evidence that the very same people become chicken and overreact in the face of real risks; the exact opposite. For the Stoics, prudence is connatural to courage, the courage to fight your own impulses.
• The courage to stand up for an idea, and enjoy death in a state of thrill, simply because the privilege of dying for truth, or standing up for one's value, had become the highest form of honor.
• You express your opinions; it can hurt others (who rely on it), yet you incur no liability. Is it fair?
• Words are dangerous: postdictors, who explain things after the fact, because they are in the business of talking, always look smarter than predictors.
• Never ask anyone for their opinion, forecast, or recommendation. Just ask them what they have, or don't have, in their portfolio.
• The psychologist Gerd Gigerenzer has a simple heuristic. Never ask the doctor what you should do. Ask him what he would do if he were in your place. You would be surprised at the difference.
• You were wrong for years, right for a moment, losing small, winning big, so vastly more successful than the other way.
• For Mother Nature, opinions and predictions don't count; surviving is what matters.
• Behind you is the sea, before you, the enemy. You are vastly outnumbered. All you have is sword and courage.
• Never put your enemy's back to the wall.
• I rely on my naturalistic and ecological instincts.
• It would be nice if he acted with some personal humility.
• You can't fool too many people for too long a period of time.
• Never trust the words of a man who is not free.
• The definition of the free man, according to Aristotle, is one who is free with his opinions, as a side effect of being free with his time.
• He is free who owns his own opinion.
• Not deriving your personal and emotional identity from your work, and viewing work as something optional, more like a hobby.
• "Absence of evidence" and "evidence of absence".
• The researcher's free option in his ability to pick whatever statistics can confirm his belief.
• The fooled-by-data effect is accelerating.
• All I want is to remove the optionality, reduce the anti fragility of some at the expense of others. It is simply via negativa.
• Everything gains or loses from volatility. Fragility is what loses from volatility and uncertainty.
• Convexity is the response by a thing that likes disorder.
• Via negativa: in action, it is a recipe for what to avoid, what not to do, substratction, not addition, say, in medicine.

Data Preparation for Machine Learning - Part I, II, III - Jason Brownlee

 Preamble

• This post is an extract of the book "Data Preparation for Machine Learning" from Jason Brownlee.

Chapter 4: Data Preparation Without Data Leakage

• Naive application of data preparation methods to the whole dataset results in data leakage that causes incorrect estimates of model performance.
• Data preparation must be prepared on the training set only in order to avoid data leakage.
• Normalize data: scale input variables to the range 0-1
• Standardization estimates the mean and standard deviation values from the domain in order to scale the variables.
• The k-fold cross-validation procedure generally gives a more reliable estimate of model performance than a train-test split.

Part III: Data Cleaning

Chapter 5: Basic Data Cleaning

• Identify Columns that contain a Single Value
• Drop Rows
• Get duplicates
• Drop duplicates

Chapter 6: Outlier Identification and Removal

• Standard Deviation Method
• 1 standard deviation from the mean covers 68% of data
• 2 standard deviations from the mean covers 95 percent of data
• 3 standard deviations from the mean covers 99,7% of data
• Three standard deviations from the mean is a common cut-off in practice for identifying outliers in a Gaussian or Gaussian-like distribution.
• Interquartile Range Method
• Not all data is normal or normal enough to treat it as being drawn from a Gaussian distribution. A good statistic for summarizing a non-Gaussian distribution sample of data is the interquartile Range or IQR. The IQR is calculated as the difference between the 75th and 35th percentiles of the data and defines the box in a box and whisker plot.
• We can calculate the percentiles of a dataset using the percentile() NumPy function.
• Automatic Outlier Detection
• Each example is assigned a scoring of how isolated or how likely it is to be outliers based on the size of its local neighborhood.
• The scikit-learn library provides an implementation of this approach in the LocalOutlierFactor class.
• Standard deviation and interquartile range are used to identify and remove outliers from a data sample.

Chapter 7: How to Mark and Remove Missing Data

• Pandas provides the dropna() function that can be used to drop either columns or rows with missing data.

Chapter 8: How to Use Statistical Imputation

• Datasets may have missing values, and this can cause problems fro many machine learning algorithms. As such, it is good practice to identify and replace missing values for each column in your input data prior to modeling your prediction task. This is called missing data imputation, or imputing for short. A popular approach for data imputation is to calculate a statistical value for each column (such as a mean) and replace all missing values for that column with the statistic.
• Missing values must be marked with NaN values and can be replaced with statistical measures to calculate the column of values.
• The scikit-learn machine learning library provides the SimpleImputer class that supports statistical imputation.
• To correctly apply statistical missing data imputation and avoid data leakage, it is required that the statistics calculated for each column are calculated on the train dataset only, then applied to the train and test sets for each fold in the dataset. This can be achieved by creating a modeling pipeline where the first step is the statistical imputation, then the second step is the model. This can be achieved using the Pipeline class.
• Testing different imputed strategies. The author proposes to test different strategies when replacing data by comparing the mean, the median, the mode (most frequent) and constant(0) strategies. 

Box and Whisper plot of different imputation strategies

• We can see that the distribution of accuracy scores for the constant strategy may be better than the other strategies.

Chapter 9: How to Use KNN Imputation

• Although any one among a range of different models can be used to predict the missing values, the k-nearest neighbor (KNN) algorithm has proven to be generally effective, often referred to as nearest neighbor imputation.
• Missing values must be marked with NaN values and can be replaced with nearest neighbor estimated values.
• An effective approach to data imputing is to use a model to predict the missing values.
• One popular technique for imputation is a K-nearest neighbor model. A new sample is imputed by finding the samples in the training set "closest" to it and averages these nearby points to fill in the value.
• The use of a KNN model to predict or fill missing values is referred to as Nearest Neighbor Imputation or KNN imputation.
• The scikit-learn machine learning library provides the KNNImputer class that supports nearest neighbor imputation.
• The KNNImputer is a data transform that is first configured based on the method used to estimate the missing values.

Chapter 10: How to Use Iterative Imputation

• Missing values must be marked with NaN values and can be replaced with iteratively estimated values.
• It is common to identify missing values in a dataset and replace them with a numeric value. This is called data imputing, or missing data imputation. One approach to imputing missing values is to use an iterative imputation model.
• It is called iterative because the process is repeated multiple times, allowing ever improved estimates of missing values to be calculated as missing values across all features are estimated. This approach may be generally referred to as fully conditional specification (FCS) or multivariate imputation by chained equations (MICE).
• Different regression algorithms can be used to estimate the missing values for each feature, although linear methods are often used for simplicity.
• The scikit-learn machine learning library provides the IterativeImputer class that supports iterative imputation.

Conclusion

• In this first three parts of the Data preparation for Machine Learning Book, we review the main data cleaning possibilities.
• The following part will be related to "Feature Selection".

94% des coureurs posent le pied d'abord sur le talon

 D'après le site Volodalen, 94% des coureurs posent d'abord le pied sur le talon.

Comment décider - Annie Duke

Préambule

• Ce post se veut un résumé des meilleurs extraits du livre de Annie Duke, "How to Decide".
  
  
  

Introduction

• Un bon outil d'aide à la decision cherche à réduire le rôle des biais cognitifs. (tel que l'excès de confiance, le biais rétrospectif ou le biais de confirmation).
• Une liste de "pour" et "contre" à tendance à amplifier le rôle des biais. 
• Evaluer si une décision est bonne ou mauvaise revient à examiner les croyances qui ont informé la décision, les différentes options, et comment le futur pourrait se dérouler en fonction des choix.
• Il faut éviter les pièges de la prise de décision en équipe, et spécialement la pensée de groupe.

Chapitre 1 : Le biais du résultat

• Le biais de résultat (resulting bias) est un raccourci que l'on prend pour juger de la qualité d'une décision en fonction de la qualité du résultat.
• L'expérience est supposée être notre meilleur professeur, mais parfois nous lions trop directement la qualité d'un résultat à la qualité de la décision.
• Prendre de meilleures décisions signifie faire une meilleure prévision des différents scénarios.
• On a tendance à remarquer lorsqu'on pas de chance et à oublier lorsque c'est un coup de bol.
  
  
  
• Le biais de résultat fait que nous manquons de compassion pour nous-même et pour les autres.
• On se flagelle lorsque les choses ne fonctionnent pas comme on l'aurait espéré.
• La chance est ce qui intervient entre ta décision et le résultat. Le biais de résultat enlève la perception de la chance.
• Quand tu prends une décision, tu peux rarement prédire un bon résultat. L'objectif est de choisir l'option qui va conduire à un ensemble de résultats potentiellement positif.
• Tu prends une décision basée sur un nombre d'informations insuffisantes.

Chapitre 2 : Le biais rétrospectif

• Afin d'évaluer la qualité d'une décision et d'en apprendre les leçons, tu as besoin d'évaluer ton état d'esprit honnêtement et te rappeler qu'est-ce que tu savais et qu'est-ce que tu ne savais pas avant la décision.
• Les faits influent sur ta mémoire de ce que tu savais avant les faits.
• Nous sommes généralement très prompts à critiquer celui qui a pris la décision, en négligeant de se mettre nous-mêmes dans la peau du décideur au moment de la décision.
• Le biais rétrospectif ("hindsight bias") est la tendance à croire que le résultat était prévisible ou inéluctable.
• Le biais rétrospectif, comme le biais du résultat, se manifeste par une influence trop importante du résultat. Le résultat vous empêche de voir ce que tu savais au moment de la décision.
• L'extension dans votre mémoire ("memory creep") est le fait que les faits envahissent votre mémoire de ce que vous saviez avant la décision.

Chapitre 3 : La décision qui ouvre des univers multiple

• La checklist décisionnelle t'aidera à qualifier la qualité de la décision en créant un arbre de décision. Pour cela il faut :
1. Identifier la décision
2. Identifier le résultat actuel
3. Avec le résultat, créer un arbre avec les autres résultats qui étaient possibles au moment de la décision.
4. Analyser les autres résultats afin de mieux comprendre qu'est-ce qu'il faut retenir du résultat actuel.
• Le résultat seul ne te permet pas de dire si une décision était bonne ou mauvaise.
• Il y a une asymétrie dans notre volonté de remettre les résultats dans leur contexte : on le fait plus facilement lorsqu'on échoue que lorsqu'on réussit.
• Tu ne peux pas complètement apprendre d'un résultat sans comprendre les autres évènements qui auraient pu se produire.
• Et si ? Un résultat possible d'une décision est un résultat qui ne sait pas produit. Une imagination.

Chapitre 4 : Préférences, bénéfices et probabilités

• Six étapes pour une meilleure prise de décision

1. Identifier le jeu des résultats, des issues possibles.
2. Identifier vos préférences par association des bénéfices pour chaque résultat, issue ou objectif. Jusqu'à quel niveau de satisfaction vous appréciez le résultat en fonction de vos valeurs.
3. Estimer la probabilité de chaque réalisation potentielle.
4. Estimer la probabilité du résultat que vous aimez et que vous n'aimez pas pour l'option considérée.
5. Répéter les étapes 1 à 4 pour les autres options.
6. Comparer les options entre elles

• Identifier l'ensemble des résultats possibles pour une prise de décision est déjà une amélioration énorme par rapport à une décision prise sur le résultat d'une décision passée.
• La poubelle d'une personne est le trésor d'une autre personne.
• Demander l'avis est un excellent outil décisionnel tant que tu es explicite sur tes objectifs et tes valeurs.
• La taille du bénéfice est importante.
• Si tu fais un investissement qui te rapporte de l'argent, c'est un gain. Si tu perds de l'argent, c'est une perte.
• Le bénéfice peut être non monétaire : bonheur, temps, amélioration personnelle, estime de soi, santé, gentillesse
• Evaluer la qualité d'une décision consiste à comprendre si les avantages sont valables par rapport aux risques des inconvénients.
• Un des plus gros obstacles à une bonne prise de décision est de croire que la décision est bonne ou mauvaise et qu'il n'y a rien entre ces deux choix.
• Une bonne prise de décision passe par la volonté de deviner.
• Tu sais forcément quelque chose, et ce quelque chose est mieux que rien.
• Qu'est-ce que je sais déjà et qui rendra mon estimation plus étayée ?
• 
  Qu'est-ce que je peux trouver qui m'aidera dans ma prise de décision ?
• Ton choix est toujours une estimation des différents scénarios possibles.
• Pour beaucoup de décisions, il y a souvent un aspect très particulier qui vous intéresse.
• Ajouter des probabilités à un arbre décisionnel va améliorer de manière significative la qualité de vos décisions par rapport à simplement identifier les possibilités à vos préférences.
• Si tu ne poses pas la question, tu n'auras pas la réponse.
• Tu peux obtenir un certain contrôle sur l'incertitude qui se cache derrière l'imperfection de l'information.
  

Chapitre 5: Le pouvoir de la précision

• Beaucoup de choses que tu ne connais pas se trouvent dans la tête d'autres personnes. Aussi obtenir du feedback des autres sur les choses que tu crois et les décisions que tu prends sera le meilleur outil pour extraire la connaissance du monde.
• Pour avoir un feedback de qualité sur tes décisions et tes croyances, tu dois parler le même langage que la personne qui te donne le feedback.
• Si tu crois qu'un évènement à 30% de chances de se passer et que tu parles avec quelqu'un qui a une information fiable que cet évènement se déroulera avec 70% de chances, cela te sera utile pour lever le désaccord.
• La précision enlève le désaccord.
• La précision est importante.
• Qu'est-ce que je sais, et qu'est-ce que je peux apprendre ? 
• Quand tu exprimes quelque chose avec certitude, les autres seront moins enclins à te contredire. Soit parce que ils pensent qu'ils ont torts et ne veulent pas s'embêter en partageant ce qu'ils pensent, ou ils pensent que tu as tort et ils ne veulent pas te mettre dans l'embarras. Ceci est un problème si tu as un rôle de leader.
• Quand tu crées une limite inférieure et une limite supérieure, tu exprimes que tu te trouves quelque part entre pas de connaissance et une connaissance parfaite. Si tu laisses comprendre à celui qui t'écoute que tu n'est pas sûr fera que cette personne sera plus enclin à te partager de l'information utile parce que tu as demandé de l'aide.
• Approche la qualité de ton information avec plus de scepticisme. Ce scepticisme te rendra plus curieux à questionner tes propres certitudes et plus enclin à chercher ce que les autres savent. Et ceci améliorera la qualité de tes décisions.
• Fais en une habitude de te questionner toi même : si je me trompais, quelle serait l'alternative ?
• Utilise la méthode du choc test pour savoir si tes limites inférieures et supérieures sont raisonnables.  Est-ce que tu serais vraiment choqué si la réponse correcte était en dehors des limites ?
• Etre trop confiant en soi contrarie la prise de décision.

Chapitre 6 : prendre des décisions avec l'aide de l'extérieur

• Ce que tu vois très clairement chez les autres est difficile de voir pour soi-même.
• Votre perspective n'est pas si bonne quand tu es au centre.
• Tes croyances créent un embouteillage pour la bonne prise de décision.
• C'est très dur pour nous de voir le monde d'un point de vue extérieur.
• Biais de confirmation : notre tendance à noter, interpréter et chercher de l'information qui confirme nos croyances.
• Biais de non confirmation : notre tendance à appliquer plus d'importance à une information qui contredit nos croyances par rapport à une information qui confirme nos croyances.
• Surconfiance : surestimation de nos compétences qui interfèrent avec notre capacité de prendre des décisions en se basant sur ces estimations.
• Biais de fréquence : la tendance à sur estimer la fréquence de certains évènements  qu'il est facile de se rappeler parce qu'ils sont vivaces ou parce que nous les avons vécus plusieurs fois.
• Biais du récent : croire que les évènements récents sont plus susceptibles de se reproduire qu'ils ne le sont réellement.
• Biais de l'illusion du contrôle : surestimation de notre capacité à contrôler les évènements. En d'autres termes, sous-estimer l'influence de la chance.  
• L'intérêt d'avoir le point de vue des autres est que même s' ils disposent des mêmes faits, ils peuvent voir les choses de manière différente.
• Si tu veux te créer un outil d'aide à la décision qui amplifie les biais , cela devrait ressembler à une list de pour et de contre.
• Si tu n'as pas une vue précise de tes compétences, tu peux prendre de très mauvaises décisions.
• C'est inconfortable de penser à l'échec, mais cela vaut le coup de vivre dans cet inconfort parce que tu seras mieux préparé si les choses ne se passent pas comme dans ton idéal.
• Les gens brillants sont meilleurs pour construire des arguments convaincants qui soutiennent leurs point de vue et qui renforcent les choses auxquelles ils croient.
• La personne que tu as le plus de chance d'induire en erreur est toi-même. Et tu ne t'en rends pas compte parce que tu le vis de l'intérieur.
• Remercie les gens qui, de bonne foi, sont en désaccord avec toi car c'est une action aimable de leur part.
• Cherche le point de vue extérieur avec un esprit ouvert. 

Chapitre 7 : casser la paralysie de l'analyse ou comment optimiser le temps de votre prise de décision 

• Le temps est une ressource limitée que tu dois dépenser avec sagesse.
• La capacité de savoir quand il faut décider rapidement, ou quand il faut ralentir est une compétence cruciale à développer.
• Tu dois être concentré sur les façons d'extraire les informations du monde afin de transformer l'univers de ce que tu ne connais pas dans quelque chose que tu connais.
• Plus tu connais tes préférences, plus il te sera facile de prendre des décisions.
• Le test du bonheur : demande toi si la décision que tu prends, bonne ou mauvaise, aura des chances significative sur ton bonheur dans un an. Si la réponse est non, la décision passe le test, ce qui signifie que tu peux prendre ta décision rapidement.
• Les choix qui se répètent aident à diminuer le regret.
• Lorsque le même type de décision revient couramment, tu as donc des chances répétées de faire ton choix, en y incluant même des choix que tu as rejeté dans le passé.
• Freeroll : une situation dans laquelle il y a une asymétrie entre les avantages et les inconvénients car les pertes potentielles sont insignifiantes.
• Le concept de freeroll est un modèle mental utile pour cibler des opportunités à saisir rapidement.
• Plus tu t'engages rapidement et moins l'opportunité à des chances de disparaître.
• L'impact de la décision obscurcit le freeroll.
• La peur de l'échec ou du rejet peut vous paralyser.
• Quelque soit l'option que je prends, quelle sera ma plus grande perte ?
• Plus tu as un grand nombre d'options, plus il y a de chances qu'une de ces options te semble intéressante. Plus tu auras des options qui te semblent intéressantes, plus tu passeras ton temps dans la paralyse de l'analyse.
• C'est le paradoxe : plus de choix signifie plus d'anxiété.
• Pour toute décision, passe ton temps à départager les choses que tu aimes de celles que tu n'aimes pas.
• Le coût de l'opportunité : quand tu choisis une option, tu perds le gain potentiel associé aux options que tu ne choisies pas.
• Plus les gains associés aux options que tu ne choisies pas sont grands, plus le coût de l'opportunité est grand.
• Si je prends cette option, qu'est-ce qu'il m'en coutera de l'abandonner ?

Chapitre 8 : la puissance de la pensée négative

• Pensez positivement, mais planifiez négativement.
• Ne confonds pas la destination avec la route.
• Si cela devait mal se passer, de quelle manière cela se passerait-il ?
• Lorsqu'il s'agit de navigation, la pensée négative vous permet d'arriver à destination de manière plus sûre.
• Le contraste mental est le processus d'imagination des obstacles qui pourraient surgir sur la route de votre destination. 
• La douleur mentale vous conduit aux gains réels.
• Le biais du statu quo : la tendance à croire que les choses d'aujourd'hui seront les mêmes demain.
• Analyse pré-mortem : tu t'imagines quelque part dans le futur après avoir échoué sur un objectif, et tu regardes derrière toi sur comment tu en es arrivé là.
• Plus tu auras une vue large et complète sur ton futur, plus facile sera ta décision.
• La valeur ajoutée de l'analyse pré-mortem pour les groupes : tu ajoutes plus de matière grise.
• De manière intuitive, nous pensons que plusieurs cerveaux sont meilleurs qu'un seul pour la prise de décision.
• Les analyses pré-mortem de groupe permettent d'exposer et encouragent les différents points de vue.
• De même que la précision se trouve à l'intersection entre la vue externe et la vue interne, la vue la plus précise du futur se trouve à l'intersection entre l'analyse pré-mortem et le "backcasting".
• La vue externe permet de discipliner les biais cognitifs qui résident dans la vue interne.
• La table d'exploration de la décision.
  
  
  
• La première chose que tu devrais considérer après un travail pré-mortem et un backcasting est de savoir si tu modifies ton objectif ou si tu changes de décision.
• Rechercher les manières de mitiger l'impact des mauvaises surprises si elles arrivent.
• De même qu'Ulysse, tu peux physiquement t'empêcher de prendre des mauvaises décisions.
• Combien de fois j'ai fait exception à la règle récemment ?
• Lors de l'occurence d'un mauvais résultat, il est difficile de voir la situation de façon rationnelle.
• Vous compromettez la qualité de votre décision lorsque votre état émotionnel est intense suite à un mauvais résultat.
• Est-ce que tu prends les choses de manière personnelle ? ou est-ce que tu entres dans la confrontation ? ou est-ce que tu utilises un langage particulier ? ou est-ce que tu t'engages dans des chemins de pensée lorsque tu es guidé par l'émotion ? Si oui alors tu es dans le mode "Tilt"

Chapitre 9 : l'hygiène de la décision. 

• Le problème lorsque vous proposez d'abord votre opinion lorsque vous demandez un conseil à quelqu'un est que cela augmente de manière significative le fait qu'il exprimera la même opinion que vous. (Salomon Asch)
• Un des meilleurs outils pour améliorer ta prise de décision est de demander le point de vue des autres. Mais tu ne peux le faire que si tu demandes leur vrai point de vue au lieu de voir votre perspective renvoyée.
• Quand vient le moment de la décision, il est plus intéressant d'avoir plus de cerveaux qu'un seul. 
• Si tu proposes d'abord ton avis, l'opinion de celui qui répond peut changer, ou la personne peut ne pas vouloir dire sa vraie pensée, ou elle peut penser qu'elle à tort.
• Il est impossible de tomber dans le piège du biais rétrospectif si tu ne connais pas le résultat.
• Si tu connais le résultat cela détruit le feedback.
• Le fait que tu ne communiques pas le résultat et tes croyances avant de demander le feedback à une personne fait que cette personne se trouve dans les mêmes conditions que toi au moment de la décision.
• Il faut solliciter le feedback de manière indépendante.
• Eliciter le feedback de manière indépendante et partager le avec le groupe avant le meeting.
• Il faut rendre anonyme le feedback.
• L'effet de halo est un biais cognitif dans lequel une impression positive sur une personne dans un domaine vous fera aussi donner une une opinion positive de cette personne sur un problème tout autre.
• La qualité de l'information que vous donnez avant la demande de feedback est très importante. Vous devrez préciser vos objectifs, ce qui pour vous à de la valeur et vos ressources.