(Get) 👈 B.o.o.k.s Algorithms to Live By

Algorithms to Live By

Link to Downlload ⇒ Downlload Now!

Link to Read ⇒ Reaad Now!

 

Review : What should we do, or leave undone, in a day or a lifetime? How much messiness should we accept? What balance of the new and familiar is the most fulfilling? These may seem like uniquely human quandaries, but they are not. Computers, like us, confront limited space and time, so computer scientists have been grappling with similar problems for decades. And the solutions they’ve found have much to teach us. In a dazzlingly interdisciplinary work, Brian Christian and Tom Griffiths show how algorithms developed for computers also untangle very human questions. They explain how to have better hunches and when to leave things to chance, how to deal with overwhelming choices and how best to connect with others. From finding a spouse to finding a parking spot, from organizing one’s inbox to peering into the future, Algorithms to Live By transforms the wisdom of computer science into strategies for human living. Read more

 

Review : Within the “Scheduling” chapter they indicate that “of the 93% of the problems that we do understand… only 9% can be solved efficiently…” This appears to apply to much of the book regarding the various topics covered. Often the treaties are interesting, but the solutions are often impractical, inapplicable, or outdated. Within the “Caching” chapter the authors make much about human memory and Media headlines both fading away very rapidly as time goes by. They feel this forgetting pattern is part of an underlying universal principle. It may be, but when you look at their own graphs (pg. 101) on the subject they omit to emphasize that the graphs have completely different scale on the x-axis. What human beings forget in a matter of hours, the Media moves on in a matter of days. It is a very different time scale that diminishes the insight associated with this principle besides the obvious: yes indeed individuals and even societies have a limited memory (on their own different respective scale). They are often not in tune with the information age. For instance the algorithm that dominates the first half of the book is the 37% rule that you should stop gathering data regarding decisions after researching 37% of the data you were considering exploring. This virtually applies to everything. If you were planning to date 10 different people before getting married in order to “shop around” you apparently have enough info after dating the first 4. If you are planning to rent an apartment the same is true (you have enough info to make a good choice after interviewing the first 4 applicants out of 10). If you are planning to sell a house you can accept an offer after passing on the first few of them. If you are recruiting and hiring a secretary the same principle holds up. However, with the online world we have so much more information than the world the author describes. Regarding mating with numerous online websites one has so much more information and choices than what the 37% rule would suggest. The same is true if you are hiring a secretary, you can just advertise on an online platform receive a 100 resumes in a few days. Filter those resumes, interview just a few candidates, select the best one and be done with it. This recruiting renders the 37% rule irrelevant (you don’t need to interview 37 candidates out of 100 since you already have a lot of info on all of them before interviewing them). Also, absent from the math the authors convey is the concept of supply and demand. When selling a house, this transaction is dominated by the local supply and demand. For instance, anyone who has sold a home during the housing crisis most probably did not have the luxury to wait out for better offers such as the 37% rule would suggest. In general, waiting for a better offer does not work well in real estate. A house number of days on the market is a measure of how stale a prospective home sale gets. Waiting for better offers (37% rule) typically does not work. That is why sellers remove their homes from the market to give them a fresh reset. Also absent from the authors’ calculations are moral considerations as they state: “if you are a skilled burglar and have a 90% chance of pulling off each robbery (and a 10% chance of losing it all [by being caught] ), then retire after 90/10 = 9 robberies. Cool math but not exactly “Algorithms to Live By” as the title suggests. On other occasions, they do not support or explain the underlying math at all. Such is the case for the Gittin Index they cover on page 39 to 42. The latter is associated with counterintuitive results that remain confounding. Other algorithms appear flawed. This includes the Upper Confidence Bound algorithm that supposedly guarantees minimal regret. I am unclear how that would be the case because by selecting such an option you also take the maximum risk. That’s what condo flippers did during the housing crisis Leveraging gets you up on the Upper Confidence Bound… but also the Lower one. The authors cover the most important subject Bayesian statistics within chapter 6. However, their treatment of the subject focuses a lot more into challenging technical considerations like the probability distribution of the a priori events (Normal, Power, Erlang, etc.) rather than on explaining the basics of Bayes theorem. Without establishing a good foundation explaining Bayes theorem any insights regarding a priori events distributions are rather obfuscating. For a better coverage of Bayesian statistics Nate Silver’s “The Signal and the Noise” is a lot more edifying. Several of their chapters’ subjects and titles use confusing play on words that make them sound like they are relevant to your daily life but they really are not. The chapter on “Relaxation” has nothing to do with relaxation. It describes mathematicians removing technical math constraints from very challenging problems in order to being able to solve them. The chapter on “Randomness” has also little to do with a layperson’s meaning of randomness. Instead, it deals with technical math concepts regarding sampling, Monte Carlo simulation, and randomized algorithms. Those represent another set of math strategies to solve what would be otherwise unresolvable problems. The book is not all bad. The chapter on “Overfitting” is excellent. That’s even though it is still aimed at the math geek crowd and provides little in terms of “Algorithms to Live By.” This book is truly very mistitled and mispecified in terms of audience target. In this chapter, they warn against developing models associated with higher degree polynomials to better fit the curve of a given data set. This is not just with higher degree polynomials but often any model that has a lot of variables that fit the history of the data really well. Such complex models with many variables often do a worse job of predicting given new data vs. much simpler models that do not fit the learning sample of the model as well. Their referring to cross-validation to test for overfitting, regularization to preempt overfitting, and stepwise methods to build streamline models consists of interesting arcane math technicalities. None of them have much relevance in your daily life decisions. The chapter on “Game Theory” is also excellent. Their treatment of Game Theory is very good. Additionally, their explanation of a specific Game Theory situation: Information Cascade is truly fascinating and for once most relevant. It explains a whole lot about group behavior, asset bubbles, and related financial crises. What others have often described as the “madness of crowds” to explain bubbles may be better explained by information cascades. During the most recent financial crisis, each relevant party may have followed their own rational economic interest. But, the whole economic sector was plagued by negative equilibria that lead to inevitable disasters. This is a characteristic of information cascades as described within the book in the section “Information Cascades: The Tragic Rationality of Bubbles.” My rating reflects that there are only two excellent chapters out of 11, and most of the math content is not really relevant to your daily life. If you have not heard of the 37% rule, there is a good reason for that; it is obsolete.