Review From User :
Tim O'Reilly, who I admit to having no awareness of prior to buying this book, has obviously had a front row seat at the birth and development of the digital economy. And he's either a prolific note taker or has a large research staff.
However it came into being, this is a thorough, if not exhaustive, review of the history of digital. At 448 pages, it is quite literally a tome of a book. And while the author is clearly a competent documentarian, I wouldn't call it a quick read. I would have accepted his references with less supporting documentation but engineers, admittedly, may be more demanding on that front.
For me, the book is really two books. The first book is all about the history of Silicon Valley and its creations. When he noted "the genius of TCP/IP" I considered putting the book down, as I don't have a clue what that is and don't really have any interest in learning as long as my Mac and Kindle work.
The Internet has also trained me in the value of "chapter learning." There is a lot I don't need to know because if and when I do I can turn to Google and YouTube. But I slogged through and it was undoubtedly good to get more informed. (We're all a little lazy on that front today.)
The second book-the one about the metaphorical Silicon Valley's place in the word-was pure gold. In this book the author takes an inquisitive scalpel to the frustrating world we now live in and, explains it, isolates some of the root causes, and offers some prescriptions.
While I am not a techie, I am a mathematician and philosopher of sorts and was fully engaged by "Part III: A World Ruled by Algorithms." Algorithms drive the digital world but are little understood by the people who use its services. An algorithm is a recursive computation that provides, particularly when used in groups, informed answers to problems like how to rank data or answer a search. A computation, however, is not a calculation in the way that 2+2=4 is; least of all when context is factored in. Algorithms will give you an answer but not necessarily "truth." That, more often than not, is a matter of perspective and your personal standard of precognitive conclusion.
Which is precisely why "fake news" will be impossible to ultimately prevent. Even Facebook's vision of communities won't help. It is community that is the problem to begin with. In the end, the news coming from the "other community" is all fake because, by definition, it is not substantiated if we are unwilling to accept that it is.
Algorithmic bias, I believe, is the biggest challenge our society and our economy faces at the moment. I dare say it is more immediate than climate change for the simple reason that the Internet has become integrated with our economy, our politics, and our culture to such a degree that if it fails our world will come tumbling down.
And it will fail, I believe, because of algorithmic bias, which will undermine trust in the Internet, or, more precisely, the Internet gatekeepers. Trust is pivotal to the Internet ecosystem and the gatekeepers, to date, have protected it with skill and determination.
The author actually lays out the argument quite well when he notes that traffic tickets handed out by intersection cameras are quite "fairly" distributed. Who can argue with the time-stamped image And he's right, of course. But what if the cameras are only installed in certain neighborhoods and not installed in certain other neighborhoods
The problem is not the algorithm per se, it is its application. The author correctly notes, "The characteristics of the training data are much more important to the result than the algorithm." Bingo. And that will be an impossible problem to fix to everyone's satisfaction. (Compromise is not exactly the ideal of the day.)
And the courts, I predict, won't help. The Digital Millennium Copyright Act of 1998 exempted ISP's from all copyright laws because they are, theoretically "neutral." This protection, O'Reilly argues, is both warranted and critical. The warranted argument is moot, however, because ISPs will eventually lose that protection in the courts. Semantics are a double-edged sword in our legal system. Our legal system turns on semantics and the distinction between a "neutral platform" and a content provider will ultimate be erased once the mobs outside of SV turn on it.
The author's solution to algorithmic bias is to double down-install more and more robust algorithms that are measured by the right results. (Google's quest for "relevance" won't do it.) And that will help. It will not, however, erase a problem that people are only now even becoming aware of. And the very psychological attributes that allow people to be hoodwinked also work in reverse. Once the tipping point is reached, convincing them that you now tell the truth is next to impossible.
In the end I couldn't agree more with O'Reilly that the real problem we face today is the master algorithm of serving the shareholder. "It's essential to get beyond the idea that the only goal of business is to make money for its shareholders." As a former CEO myself, he is absolutely right; we have hollowed out our economy and our souls and given it all to management and their investors, who now enjoy a very outsized portion of our miraculous economic output. And we are destroying our economic future in the process.
"People have a deep hunger for idealism," O'Reilly notes. And I agree. We can survive, or, if we don't survive in the short term, dig our way out. Our resilience is legendary.
I further agree with O'Reilly that the concerns about the robots putting us out of work are overstated. There will always be plenty to do.
Fixing algorithmic bias, however, will be painful. Some wealth will be lost. Some power will have to be redistributed. It won't happen without a battle. Bravo to Tim O'Reilly, however, for putting this very important topic on the table for discussion.
This will, I believe, prove to be a seminal book on a topic of truly epic importance.
Media Size : 1.3 MB