Book Review: "Accelerate: Building and Scaling High Performing Technology Organizations", by Nichole Forsgren, Jez Humble, and Gene Kim

I got this book on the recommendation of my manager; I had it for a few months now in my Amazon "Saved for Later" (seriously, what's up with that, if I add an item to my cart and then don't want it I can only save for later vs. adding to my wishlist), and during a 1:1 I mentioned I had it there, and he bought it for me!! So I spent the past month or so reading through it on and off.

I think this book confirmed a lot of my original ideas about how software in production should be working. I say original as coming out of college into my first job and reading through a bunch of books on software. The thing about "original" though is I gradually lost my compass as the years went on, and doubted my original beliefs because of places I've worked at had different needs (and honestly addressed them in ways that could have gone better). This book brought back a lot of the core arguments, and backed them up with data.

I think what this book really excels at is selling the concept of survey management and qualitative data. I think for me as an engineer, I tend to value quantitative data a lot. The arguments against quantitative data they made, like "oh if something went wrong with your quantitative data a whole bunch of data points are off in the same way vs. much more stochastic outliers with bad qualitative data" and "you cant capture everything with your quantitative data" didn't really appeal a whole lot to me. Yeah, quantitative data can be bad, but there are ways in order to preserve data quality and I think at this point in the tech world people should be on the same page regarding quantitative data not capturing the full source of truth. I do agree though that having more data is always nice, and given data around people's thoughts, feelings, perceptions, etc. changes constantly, taking snapshots of that data and evolving that schema with people trained in-house to do so is really important for measuring org efficacy. I've seen this at my current place; we use this tool called Lattice, and it's really nice to keep track of 1:! / skip-level agenda items, weekly updates, and pulse surveys. This is on top of Google Forms (which I think come with our G Suite account), and third-party form builders like Qualtrics or SurveyMonkey.

Another good thing the book does is to tie in org culture with DevOps. I've seen this first-hand, working at orgs just with access to CI and without access to CI; it's a complete world of difference, because with CI the nits don't build up and cause friction between people and orgs. Wide permissions and audit trails help create a high-trust (but verifiable) environment, and logging + quality deployment topologies means easy tracing and patching of issues. Implementing all this really does add a smile to your lips and a bounce to your step when you're coming into work (or opening the laptop lid) every day.

One thing I wish the book did better was present its captured data. It mentioned that it's based off of a "State of DevOps" report, but I don't think the report is highlighted (or if it is, highlighted in color because this is black/white paperback). It'd also be nice in order to have a copy of the raw survey data as a CSV file hosted somewhere, so people / organizations can make their own deductions and see how the deductions are broken down within the book. There's some charts in the book to back some arguments, but I wish they were more substantive.

Overall, I think it's a quality book in order to adjust your org outlook, esp. if you're part of a larger organization (I don't know what a change assessment board is, I'm guessing it's relevant in larger orgs). Pretty quick, and fun too.