I have seen this book recommended for years by Silicon Valley moguls. Andy Grove wrote this book as he was the CEO of Intel at the most successful time of the company.

The book is holding a lot of substance, even if I disliked some of the lack of humanism.

Especially two points:

• that you should motivate your workforce by making them compete amongst each other (there is such an example where they give a score to cleaning teams, so that they start trying to beat each others’ score). This feels manipulative.
• some point about the fact that being nice is not the goal, that efficiency is the only goal. I am inclined not to believe that and feel that a company having such values is doomed to be a sad place (and ultimately, not very efficient). I hope that I am not wearing too rosy glasses.
  

Other than that, I found many other points interesting:

• that there is this idea that you promote incompetent people up to the first level where they are incompetent, and that it is a bad practice. Andy Grove makes the point that there is no way out of this. If you are promoting incompetent people instead, you are just sending a bad message to the organisation.
• There is an interesting section about interviewing. It says that it should be relatively free form, but also that there is no magic trick. It is a hard process and mistakes happen there. The conversation should not be adversarial, but open. This part is hard to summarise because there are too much good ideas, actually. Read it.
• Manager should be the teachers of their organisation: they have two levers for managing, motivating people and teaching them. The second lever is insufficiently used.
• Meetings are not evil, they are one of the tools to transfer information. Obviously, that does not mean that too much meetings are not bad or that meetings should not be well organised.
  
• That dual reporting is the normal way in larger organisations: there should be at least a layer of the organisation that reports both to mission oriented people (the salesforce, for example) and to tech oriented people (to share information and resources amongst pairs). Anything that does only one of these reporting will be either inefficient (think a sales team hiring its tech team alone) or too detached from reality (think a tech team barely speaking to a sales team. This is often called working in silos).

All in all, a good, interesting read. I felt sometimes that it could have been shorter, but the information ratio is still quite good in my opinion.