Category: Book Review

This “book” is actually a set of conversations recorded by John Willis and Gene Kim. The main thing I can offer by way of advice is; absolutely do NOT buy a printed version, this isn’t really a book. The book is apparently a inspired by a similar “Beyond the Goal” lecture done by Eliyahu Goldratt who authored “The Goal” the novel about Lean that inspired Kim to write a novel to explain DevOps. I highly recommend getting the book, even if you only listen to the modules on Lean and Safety Culture.

The first two chapters are not the most interesting. They cover Goldratt and Deming. They aren’t uninteresting, but it was more detail than I needed on the individuals.

The next two chapters are fantastic and go through a lot of the concepts that they borrowed from when creating DevOps. Goldratt and Deming as the people who they borrowed from most thoroughly when coming up with some of the DevOps concepts. They then turn to Lean and Safety Culture, covering the history of each discipline and the most important parts that were borrowed. I learned a TON from these sections. I particularly enjoyed the section on the Andon Chord (which is intended to be pulled on a production line if anything goes wrong). I had of course glanced at the concept before, but had always dismissed it as something that was only really useful for the most mature organizations. Kim and Willis exposed me to how Toyota actually deals with it. Learning that they actually panic when it’s not pulled enough and that not every pull stops everything, it’s more of an escalation. The comparisons to a pipeline are obvious. The next chapter is actually a recording of a conversation between leaders in Lean, Safety Culture, and DevOps there were a few good nuggets, but I generally would say you can skip it.

The only remaining module that I found interesting was the case studies. Particularly the stories about Target and Nordstrom. It also closes with an interesting observation that Willis had with the CFO of Goldman Sachs. Apparently he is tracking roll outs of DevOps technologies specifically by name (Kafka is mentioned specifically). It says something about critical agility is becoming, independent of the business functionality that’s being built on top of it.

The book bills itself as being for the CEO that’s looking to make sense of how the changes in IT should change the way they view the CIO and IT. It mostly lives up to that, though you’ll need to have some knowledge of agile/cloud. As a technologist, I also found it valuable to hear the language and examples the author uses because I think they’ll be valuable as I talk to development managers and infrastructure executives who are trying to figure out how to sell the agile/devops/product/team approach to a business that is used to thinking about “keep the lights on”/multi-year initiatives/chargeback. Overall, I’d recommend the book as a quick read. Mostly to remind you of things you already know/feel and to give you words that’ll help you say them.

A few things I liked in particular:

• One of the primary points that he makes is that the world is changing so quickly right now that the biggest risk facing most companies is that they won’t be able to change fast enough. With that in mind he argues that DevOps and Cloud are actually investments in Risk Management. I find this powerful for two reasons:
  1. In many companies it’s the people carrying the title “Risk Manager” who want to stop things like deployments during business hours and hosting data outside the data center. This points out that (often, not always) those risk managers are protecting the company from the wrong/lesser risk of small outages at the expense of the big risk (not being able to innovate as fast as your competitors).
  2. It helps justify the cost of some of the DevOps Transformations and Cloud Transformations that need to happen. Often these are hard to justify when they’re compared with the opportunity to deliver on business requirements or the opportunity to use automation to save money in infrastructure. Framing DevOps as a risk management play, helps explain why there’s a need to invest in making the company better able to change.
• He actively fights the urge to make “IT as a Business”. Arguing that it needs to be part of the company like finance or marketing. He, rightly, points out that in most companies IT is expected to operate like a sort of sub-contractor but not given the ability to hire freely and set salaries optimally, IT can’t choose to avoid business customers that become untenable, it can’t charge more for higher demand items instead of passing on costs (at least in most chargeback models I’ve seen). Additionally, making IT a sub-contractor means adding red tape to the change process. IT is inscentivized to offer standard products, to require the business “pay” for changing requirements, etc…
• He uses one of my favorite business people analogies about agile software development vs project based development. That it is like buying options. Business people understand the risk of buying assets vs buying options… most even remember that moment of surprise in business school when you realized just how different the price is between an option and an asset. Agile software development is like that. You can build an MVP for the smallest slice of the market you can think of and then wait and see if they like it before agreeing to pay for the rest of the project.

This book has a clever punch line… “What can we as humans learn from computers about how to approach problems that face us in every day life? Get the answer from a psychologist and a computer scientist.” A clever side effect is explaining some of the more interesting concepts in computer science (search, caching, communicating over an imperfect network, etc…) in the lens of every day life. The result is a book that not only offers clever antictdotes on every day life, it also educates us on concepts that effect computers, networks, and computer science.

Some of the interesting points that we can learn from computer science and apply to every day life:

1. I think their overall point is that there are a great deal of problems that, even with a super computer, cannot be solved unequivocally. The key is to pick an algorithm that, within an appropriate amount of time, has the best chance of giving you the best answer. Once you have confidence you’ve done that, live without regret. The authors quote Bertrand Russell on this, “it would seem we must take account of probability in judging of objective rightness… The objectively right act is the one which will probably be most fortunate. I shall define this as the wisest act.” We can hope to be fortunate- but we should strive to be wise. There are a bunch of examples that fall in to this, but my favorite is:
   • If you are approaching a problem where you have 2 months to find an apartment, you don’t feel you can really compare two apartments without seeing both of them, and you run a very high risk of losing an apartment if you don’t make an offer the minute you see it. Then it is optimal to spend 37% of your time looking but unwilling to commit, and 63% of your time ready to pounce if you find the right place. There’s still only a 37% chance you actually end up in the best place you could have with that… but you will have done the “wise” thing.
2. There’s a great little lesson on exponential backoff that I loved. Essentially, exponential backoff is a practice of making repeated failures drastically reduce the frequency with which you make an attempt. It’s common in computer science because computers can be so persistent if told to just retry infinitely. My desktop application, for example, may try to communicate with the server and fail. It then retries immediately (with a computer that a millisecond later). If both fail it will wait a whole second (perhaps even doing some other task in the mean time) before trying again. Then wait longer and longer between attempts because each time it fails the next time is less likely. I had never really thought about it, but while exponential backoff is common in computer science, people much more commonly give something a certain number of tries in rapid succession and then give up forever. The authors give this example:
   • “A friend of ours recently mused about a childhood companion who had a disconcerting habit of flaking on social plans. What to do? Deciding once and for all that she’d finally had enough and giving up entirely on the relationship seemed arbitrary and severe, but continuing to persist in perpetual rescheduling seemed naive, liable to leed to an endless amount of disappointment and wasted time. Solution: Exponential Backoff on the invitation rate. Try to reschedule in a week, then two, then four, then eight. The rate of retry goes toward zero-yet you never have to give up entirely” [maybe they were just going through something personal].
3. They talk about how computationally intensive it is to do a full sort. If you’re a computer nerd… you’ll love that they actually use big O notation to do so!!! While sorting makes searching easier, you have to do a LOT of searching to justify a full sort of a large set. Consequently they recommend a life filled with bucket sorts, where you just put things that are categorically alike together and then search through the categorical subset when you need a specific item. As a slob that treats his inbox as a steady stream… this seems to justify my actions!
4. There’s discussion of caching and different algorithms that computers use to decide what information to keep “close at hand” and what to file away for slow retrieval if/when needed; then brilliantly analogizes that to organizing your closet, filing cabinet, and house. That also justifies a certain amount of apparent disorganization!
5. They end up basically justifying agile product/project management by pointing out the level of complexity in predicting things that are years away and have increasingly high numbers of factors influencing them.

As I said though, it’s not all about learning life lessons. If you don’t know much about computers you can expect to learn how the internet communicates messages over unreliable wires (or even thin air), how computers calculate the route between two locations, and the magic of Big O notation.

I had extremely high hopes for this book after listening to an interview with the author. Unfortunately, most of what I liked about the book I heard in that 20 minute interview. I believe that the core theme of this book is a topic that will shape the next 30 years of American life. Unfortunately the vast majority of the book is spent detailing (and at times exaggerating) how evil the current system is instead of charting a course for correction without simply stopping progress. With that in mind, I’m going to offer in this review a few things I like and then discuss a few places where she’s gone too far or failed to cover a topic.

I love three aspects of the way the problem is framed:

1. I think it is brilliant that Zuboff connected Google’s (and then the rest of the Internet’s) switch from a “pay per impression” to “pay per click” model. This means that Google is taking responsibility not only for displaying someone’s ad… but getting the user to click on it. This incentivizes them to be actively matching customers and products. This was the first step on to our present slippery slope.
2. I love the comparison of being able to commoditize and sell the ability to predict what we will click on to the ability to commoditize and sell labor which began in the industrial revolution. Now, instead of needing 25 cobblers to make enough shoes for my city, I needed 25 laborers to operate the shoe factory. This gave too much power to the capitalists who owned the factory and ended up resulting in unions (and eventually safety and wage related regulations).
3. Zuboff points out on page 192 that “Demanding Privacy from surveillance capitalists or lobbying for an end to commercial surveillance on the internet is like asking Henry Ford to make each Model T by hand or asking a giraffe to shorten its neck. Such demands are existential threats.” I agree, progress/evolution cannot be simply reversed.

My problem with the book is how evil she claims the surveillance capitalists already are. Though I credit Zuboff with the comparison to the labor market (#2 above) she spends far more time comparing it to the way totalitarian leaders take advantage of their people or the way Europeans took advantage of Native Americans. Zuboff uses hundreds of pages trying to analogize surveillance capitalism to various evil periods in history, leaving very little of the book for the laws, regulations, social protests, etc… that might be necessary to cause a course correction. Perhaps she’s already planning a sequel?

Overall, I think the book is a worthwhile read because of how it brilliantly identifies one of the biggest problems of our time. Even if you will have to glaze over the sections that over-elaborate problem and deal with the disappointment of a lack of resolutions.

I was inspired to read this book by an interview with Botsman on a podcast I listen to. It’s based on a compelling overall narrative… how have we reached a point where no one trusts the President of the United States, but people agree to stay in spare rooms of complete strangers on AirBnB and will literally put their life in the hands of a self-driving car? Botsman came to fame writing about the sharing economy, so it makes sense to hear her perspective on the topic.

The book isn’t as bold as I would have liked in predicting the future. In books like these, I prefer to hear an experts opinion on what is likely to happen, whether it is good or not, and what that means we could/should be aware of. Botsman prefers to focus only on the third of these topics; the dangerous trends we are working towards. That said, she does offer a couple compelling cliffs that we appear to be confidently striding towards:

1. Not Evaluating Bot’s Motives – Botsman weaves together stories of her daughter interacting with Alexa and scientific studies to show that as people become more comfortable with technology, we are too trustworthy of its “answers”. Her concern is that these bots that we “trust” are not there as a public service, but to influence us in to certain behaviors.
2. Trusting Technology Before It’s Ready – Botsman also points out that there are a LOT of areas that artificial intelligence and algorithms have not figured out how to do well, but we seem to be willing to trust them without testing them. There are several good academic studies where this behavior is exhibited.
3. Algorithmic Trust (Blockchain) Can Be Manipulated – Botsman covers the Ethereum DAO hack very well. If you don’t know much about Block Chain you’ll be able to follow it. She points out two concerns with the hack. First, that the hacker didn’t really “break” the system, just found a way to use it that was unintended… imagine if Ethereum had become an international currency before that happened. Second, that a relatively small group of people decided to effectively “undo” the change. In this case they were doing “good”, but it still served to demonstrate that Blockchains can be manipulated.
4. Reputation Based Trust Getting Out of Hand – The topic of the Social Credit Score in China is covered in more detail than I’ve ever read before and it is terrifying! It’s essentially a credit score system that doesn’t just evaluate your wealth and default history, but you friends, your political leanings, the places you’ve gone, your grades in school, your hobbies, and more. Essentially this will give the government access to incentivize all manner of behavior. The consequences will be real too, “people with low ratings will have slower internet connectivity; restricted access to more desirable night clubs, restaurants, and golf courses. To quote a government document, “allow the trustworthy to roam every where under heaven while making it hard for the discredited to take a single step.” While it is easy to say, that’s just China… we already evaluate people based on the number of followers they have, little blue check marks, star ratings on AirBnB or Uber, etc…

Overall, I recommend the book, but only if you are interested in these topics and would like a little hand book of the current state of them. I also don’t think this book will age well as these topics are evolving quickly. If you happen to come across this blog post after 2019… I doubt I’d still recommend it.