In many ways this book contrasts good and evil in human political organization - and that requires some definitions: an understanding of what is "good" and what is "evil".
Something is considered Good in this book if it increases the freedom individual members of the society within which the action takes place have to make and implement their own decisions, Evil if it reduces it. Thus religions and political systems based on valuing the right of self-determination for others serve Good, while those valuing hierarchical control over individual freedoms serve Evil.
Good and Evil are absolutes, but their application requires first changing the terminology to one of "doing good", or "doing harm", and then trying to weigh one set of choices against another. Thus killing or enslaving someone is always bad, but doing so in defense of the right of self-determination can net out good - so capital punishment and incarceration are both plainly evil, but killing a violent criminal during the commission of a crime, or jailing someone who has committed one or more crimes and will do so again if given a chance, usually does more good than harm.
The principle applies to social issues across the board: something is good if, on net, it expands the range of choices other people can make, bad (or evil) otherwise. Intent is never the issue here, reality is: bad programs embarked on with good intentions are bad programs - the ends justify the means only if the entire process, from beginning to end, produces more good than it does harm.
Affirmative action in American education provides a nice example of this. Although largely undertaken with good intentions and widely considered defensible on that basis, colleges, universities, and even high schools forced to cater to students who do not meet pre-affirmative action academic standards have responded either (or both) by forcing the unqualified to fail in competition for a brass ring they can't reach, or by adding support services while dumbing down their programs - thereby reducing educational opportunities for high achievers while generally devaluing all forms of education; raising its cost from both an individual and a societal perspective; denigrating the achievements of past graduates; and, generally transforming most of what happens in the non science, non professional, faculties into narcissistic exercises in political correctness.
An important digression on the use of language in the service of evil
Determining whether an action or policy is net good or evil can be very difficult - and in many cases differing people can reasonably arrive at very different judgments. One key indicator, however, is the use of language because good usually speaks clearly while evil and the perversion of language go together like salt and seawater - everyone knows, for example, that people's democratic republics are never any of those things.
Clarity of expression is not, however, even remotely a sufficient test. People can, for example, be supremely inarticulate and yet honest while writers at movement journals like The Economist are generally very good at using language and logical exposition in the service of misleading their audience - that particular magazine, for example, has lavished praise on people like Mao, Stalin, Hitler, Castro, Mugambe, and Obama while denigrating Reagan, Thatcher, Trump, and many others down to bit players like Canada's Steven Harper.
In fact, the perversion of language to cover evil is more often about fooling the already committed than about fooling others and so the most common characteristic of the dishonest use of language is the speaker's (or writer's) silent assumption that previous distortions constitute a sufficient shared knowledge base to enable clear within group communication. In its most brutal forms this simply leverages assumptions the audience is imagined to share: thus Hitler' 1939 rants assume his audience already knows that the English are conspiring with German jews and French bankers to suppress Germany's national destiny, while today's anti-Trump journalists (like yesterday's anti-Bush commentators) assume their audience knows him to be shallow, stupid, and corrupt.
In its somewhat more subtle form people who claim to argue facts simply refuse to accept anything that strays from whatever claims they personally subscribe to, demand endless documentation and proof for contrary positions, and accept only citations from media or personalities within their own community.
The most subtle, and to some extent most common, use of language to mislead in the service of evil involves using the shared assumptions underlying the meaning of words to make them mean their own opposites - the right is wrong, big is small, defeat is victory business of Orwell's 1984.
Consider, for example, that tolerance does not apply to equals - when your mother told you she would not tolerate more of that behavior she was using the word correctly because a clear guardian-dependent relationship existed, she was acting in your best interests, and she had both the right and the responsibility to enforce her views. It's not true in all cases, of course, but it is generally true that liberals who proudly pronounce themselves tolerant of other races, other creeds, other cultures, are implicitly assuming that the person or group they are speaking to accepts the idea that the speaker plays a parental role in a dependency relationship and is correspondingly willing to see whatever they're being exhorted to be tolerant of as separate, distinct, inferior, and in need of guidance.
There's an Antiques Roadshow appraisal that you may have to see twice to understand just how effectively linguistic convention can be used to deceive both speaker and audience. The work at issue in the piece, a water color by Alice Huger-Smith, depicts the unity of creation in the form of an older black nanny dozing off in the sun while pretending to fish - but the appraiser doesn't seem to see that the woman completes the picture; that this painting, to the extent that it's about anything, is about man's place at the top of God's food chain. Instead, she reports that unnamed others have described Huger-Smith's work as "mysterious" and "atmospheric" while herself describing the artist as someone "who went out into nature" to make sketches she later turned into paintings. I have no idea what the appraiser's personal views are, but the broadcast bit contains four things that are practically diagnostic for modern liberalism: a failure to see past the nanny's role and skin color to the human within; an absolute inability to see the obvious; the assumption that nature is something humans go into, but are not part of; and, the speaker's automatic recourse to wording distancing herself from the work, from the artist, and from the interpretation she offers.