It can't tell you what objectives are important nor what risks are acceptable. Those are both transitive properties: something is important to some people, something might be acceptable to some people.
There's no objective morality and no objective politics. Even if we try to limit discussion to "facts", things fall apart fairly rapidly once we start talking stochastic events and statistical morality. Controversial subjects get bogged down with duelling sets of statistics each making subtly different claims about causation.
One of the useful things about legal training is that it forces you to separate facts from law. Facts may be true or false but only law and ethics can make claims about what is right and wrong.
It should be noted that this is a controversial view, albeit one commonly held.
There are many people who believe, for example, that given only the laws of physics and the information describing the arrangement of particles making up a kitten, that they could in principle derive the conclusion that it is wrong to set the kitten on fire, without needing any further context. There are many ways to go about this.
Discerning which facts and therefore which course of action to take often involves a judgement call, a portion of which is attempting to know all applicable parties as individuals.
The portion in the original journal entry about filter bubbles and excising dissent from our daily experience speaks to this.
We cannot make wise decisions without knowing as much as possible about both those who disagree with us as well as those who agree with us. Because sometimes the "why" or "how" is as intellectually valuable as the "what".
(Not always, but more than none by a bunch!)