Only in the minds of "yay future!" sorts of people. My parents think self-publishing a non-fiction book is sketchy, and they have every valid reason for believing this. I agree with them, in fact.

Take the "outside view": if someone handed you a book and said, "This describes the most important scientific and moral challenge of our times", and you had never heard of MIRI/FHI/CFAR/LW in any way whatsoever, what are the questions you're going to ask? They'll almost definitely be credibility questions; you want to read a book that's been edited well and comes from some kind of well-known intellectual expert on the topic being discussed, preferably one from very credible institutions like academia and government. You're a non-expert, after all, and it's not worth your time becoming an expert on some obscure, highly technical topic just to judge whether you should spend $30 on a book.

Luckily, information above indicates that FHI is publishing Superintelligence under the Oxford University label, with actual academic credentials and such, in hardback print. I'll be waiting for that one before I recommend anything to "the normals".

Traditional publishing is close to becoming an atavism, a vestigial remnant of the pre-digital age.

Digital or not, the fact that an article or book has passed the filter of a reputable publisher provides useful evidence in estimating its quality.

That's most obvious for scientific publishing, where publishers employ expert editors and peer review. Publication of material intended for non-technical audiences has to meet lower standards, but there is still some level of editorial oversight.