Writing a review for Dialogue, an academic journal....this should be coming up in the next two months or so. An enlightening piece of writing.
DOUG

There are many ironies in cyberspace. One is that it promised to bring people together from all over the world because they had the same interests, passions and positions. Today, we damn the internet for bringing people together from all over the world with the same interests, passions and positions. They’ve turned into intransigent blocks of counterfactual opinion and hate. A whole universe of irony can be found in Meganets, by David Auerbach. He shows clearly, cogently and even patiently that social media and other mass data accumulators form meganets that are well beyond the control of their makers, and most importantly, that influencing them requires ironic counter-intuitive tactics.


The metaverse: everything old is new again


The book begins with a detailed look at the metaverse by Facebook, now called Meta. Meta says its metaverse is the future of the internet, a virtual world where players can purchase goods and services, and live a life virtually. Everything will be known about the players, every transaction, connection, conversation, movement, time spent on every little thing and so on, as they live out entire lives there. This will have appeal to advertisers, they say. So if you’re looking to furnish an office in the metaverse, Meta can bombard you with designers, contractors and furniture suppliers.


But the metaverse has been available for a long time. It has its fans, but it has never soared in popularity. In fact, all Meta is apparently doing is collecting every aspect of virtual life games and enclosing them in its metaverse. It is a life game in a walled garden, where all the revenues accrue to Meta. And you never have to leave to find anything you decide you want. Just like AOL used to be: a walled-off sub-internet within the open internet. Keep them inside the walls and provide for their every need and desire so you can capture every expenditure. Big deal. We’ve seen that movie before. This time you need to use a 3D mask. Made by a company Meta bought for this purpose.


What Auerbach says they’ve got right is that everything is a game. People will spend hours and hours, and crazy money on games. Meta learned this from Farmville, which Auerbach examines in depth. Make life into a new kind of game, and facebookers, a couple of billion people, will flock there. Or will they? “This time it’s different” is not the best argument.


Crypto, the lumpy oil that lubricates the internet


Auerbach then transitions to cryptocurrencies, the currency of the gaming and the metaverse, and another great irony. The whole selling point of crypto is that it is decentralized. No one controls it; it runs by rules in algorithms. There is no direction from central bankers or other suits. It is stable, self-sufficient and laughs at inflation. Everything is secure and traceable and no one can cheat.


Except hackers, who have stolen billions of dollars’ worth of crypto from individual account holders, or issued billions of new coins to themselves. In order to prevent them getting away with it, the coin has to rewrite its rules, and unfortunately, get a majority of coin holders to agree. As with anything, this is death-defyingly difficult. In one case, the bad guys owned the majority of coins, for example. And if the plan succeeds, the coin forks, so there are two valid blockchain streams, before the scam, and after, with the scammers excised out from in between. They are written out of history. This is hardly an elegant way to run a currency and points to the screaming need for - a centralized power. Someone needs to implement police actions and policy changes before the blockchain makes them permanent and irreversible. Decentralization may be its own worst enemy.


Auerbach examines the Aadhaar ID system in India, by which all government services become accessible with an Aadhaar-issued number. This is has simplified life for a billion people, no small accomplishment. But its success has led the private sector to use those same IDs for its services too, making an already gigantic system into a Leviathan with questionable privacy or recourse. Identity theft becomes well worth the risk; it’s all in that one number. The USA went through this problem with social security numbers. Everyone required them for everything. But criminal and privacy issues meant forcing the private sector to drop that practice. Even Medicare went through a laborious and expensive process to switch everyone over to a new alpha-numeric ID to get away from SSNs. This gets us closer to the concept of meganets.


What is a meganet?


Meganets are systems like metverse, facebook and Aadhaar that thrive on data. They never have enough. There’s always something new to learn. They gobble it up and spit it back in forms often totally surprising and innovative. The more data there is, the more the meganet grows new branches. The algorithms themselves run it, invent new uses and applications, and direct what users see in their accounts and feeds. As well as which ads to show them and when, and who they should follow.


The people who created them do not and cannot control them. When Google says it cannot instruct Search to forgo a certain result requested by offended groups, that is more or less the truth. They have to cripple some functionality across the board to prevent the one search result. Like a cryptocurrency forking, it is inelegant to have to force ad hoc change. Meganets are designed to learn, adapt and expand on their own. They report to no one and need no managing. Auerbach says “A meganet is a persistent, evolving and opaque network that controls how we see the world.”


Meganets are now everywhere, because so very many internet services have the critical mass to take advantage of them. Because, Auerbach says, we now produce more data every day than was produced in the history of mankind up to 2000 AD. Every day. Meganets digest it, organize the data, label and tag it, manipulate it and produce insights that are unprecedented, if only because the amount of data is so gigantic that humans are not capable of achieving the same insights. It is just not humanly possible.
The meganet universe moves and evolves so fast, humans can’t even conduct replicable experiments with them as they “are never in the same place twice,” he says.


Meganets dispense with central control. Instead they are powered by volume (of inputs), velocity (of activity), and virality (the spread of inputs by users). These are the key factors common to meganets. They give them their power. The more these factors are present and grow, the more the meganet grows, adapts, and learns. And changes what users are allowed see and do. This is the kernel of what might turn out to be a unified theory of big data and meganets.


Artificial Intelligence: the same but different


If artificial intelligence is added to the mix, the results could be anything. But Auerbach argues that AI is not anything like human intelligence. He spends a great deal of space explaining that AI is not creating a human brain. AI systems crunch data. They can pull needed data out of huge databases and discern patterns from it all. With enough data, they can determine subtle differences. Or tone and attitude. AI can write an essay with the expertise it has taught itself. Or a novel. Or create a graphic or a deepfake video. But it cannot think. He says AI has no idea what it has written actually means. The only thing that matters to an AI app is the math. Meaning has no place. Motive has no place. Purpose has no place. So while AI is a threat, it is not the kind of threat most feared in the media on a daily basis.


Writers spend a lot of time comparing AI to the human brain. People like Auerbach say it is insulting to both. AI has sorting and calculating powers well beyond any brain. And the brain can think, assign meaning and judge data that AI cannot do at all. Taking this analogy to the rest of the human body, Auerbach says writing a program to control the feedback loop on meganets “becomes like trying to write a program to administer the human body. There are too many interlocking pieces, dynamically affecting one another.” And the meganet “sprouts new organs even as its existing ones mutate.”


He says both AI and meganets work on parallelism, through which they can rack up an overall picture from the data processed separately, creating “a whole greater than the sum of its parts.” But it is still not intelligence: “It is far easier for algorithms and AI to police tone than (to) understand content.”


Although there is no center from which to control them, humans are forever trying to control meganets with negative limitations. The only response the coders have come up with is banning: banning keywords, banning accounts, banning groups. It has never worked, Auerbach says. He uses the term Whac-a-mole numerous times to describe the pointless ways people attempt to corral meganets. It has proven a worthless tactic. But it’s all they seem to know to try.


Defending against meganets: offense is the best defense


Auerbach’s outside the box thinking leads him in a counter-intuitive direction. He thinks the only way to control meganets is to dilute them with even more data. In his words: “it is only widescale and untargeted action that has a chance of changing the overall structure and function.” It’s the final irony: using big data, even made up data, to slow the data crunching we sought so hard to devise.


Here are some “laws” Auerbach has learned about meganets:
“It is far easier to put information into the meganet than to remove information from it.
It is far easier for information to spread across meganets than for it to be contained.
It is far easier for information to be wrong than it is to be right.”


So the prejudice is precisely against reining it in. Meganets and AI are meant to be free and free-roaming, by design. So why do managers persist in trying to corral meganets by breaking them? Auerbach sees another way. In his Conclusions, he proposes totally counter-intuitive solutions (that might or might not work as planned, he admits). They are based on the three principles of meganets (above): volume, velocity and virality. This kind of thinking alone makes the whole book worthwhile:

- “Cultivating instability in the meganet, rather than one’s preferred form of stability, will shatter the hardened plaque of toxicity that attaches to its worse places.”
- One form of instability is a slowdown in velocity, making it more cumbersome to circulate a meme by posting it to everyone at once.
- Groups could be limited in size to prevent them gaining critical mass.
- Posts with links could be deprioritized.
- Turn taking means prioritizing posts from those who haven’t been heard from as often. Perhaps a blockchain would demonstrate the fairness of it, rather than have people screaming they’ve been barred.
- Chaos injection means adding an unpredictable layer. Auerbach cites ranked choice voting, by which voters signal their second and third choices for second and third rounds if the first round is not conclusive. The results could be anything. It makes political microtargeting essentially impossible, for example.
- Shaking up the data means making the meganet’s learned assumptions about how various groups operate, invalid.
- Poisoning the well with inaccurate data. Or not filling in data completely or truthfully.


This is not a battle plan to win a war. This is guerilla warfare to slow down the enemy’s advance.


The future is arguable


Looking forward, Auerbach argues that meganets will divide people into ever more specialized groups, such that their leaders may be completely unknown outside their groups, as opposed to today when these leaders have global influencer rankings. I totally disagree with his scenario, based purely on what we see happening when extreme specialization takes hold. A couple of examples: Truth Social, Donald Trump’s answer to twitter, is an empty canyon, echoing very little, precisely because everyone there came to it because they agree. There’s no one to hate there, no one to troll, no one to threaten or dox. There isn’t even anyone to convert. It has no cross fertilization capabilities.


Similarly, there is an extreme-right dating app that attracts thousands of men. Unfortunately, that’s all it attracts. There are basically no women on the app for them to find a kindred spirit. Extreme-right tends to be a patriarchal, white supremacist ethos of no great interest to women. In other words, humans are not two dimensional, and won’t be labeled as easily as Auerbach and meganets seem to say. I for one cannot picture the future broken into tiny cells of same-interest people who have no desire to be seen or heard by a wider audience, or who only have one interest in life. Fame is the name of the game. If not to be, at least to rub shoulders with. Writing responses right on the posts of Barack Obama or Elon Musk is an unhoped for ability that will not be let go so easily.


David Auerbach has the gift of perspective. Even as a young man working at Microsoft and then Google, he could see the fault lines, the way the two companies differed in leveraging the internet, and how the future would unfold under each of their leaderships. He could see where they fit in the scheme of things (and therefore left one for the other then the second one too). His writing is clear, linear and simple. He explains everything in a smooth, confident and effective fashion. His arguments are measured and sound. He has researched far, wide and especially deep. He is believable and credible. It is a rare privilege to read such a fog-clearing book as Meganets, since its subject is so totally fogbound that it has led all the media and experts astray for years. It is hugely important to understanding the future of cyberspace, business, privacy, and mankind itself.



David Wineberg

A Needed Conversation. As someone also in tech at a megacorporation (though to be clear, not the same ones Auerbach has worked for) that openly seeks to employ several of the technologies discussed in this book, and as someone who finished this book right as Elon Musk's takeover of Twitter was being completed and Facebook announced that it was open to colluding with Twitter regarding content moderation... this was an absolutely fascinating look at my field and where at least one part of it currently is. But it is also written in a very approachable manner, one such that anyone who so much as uses any social media even casually or who interacts with their government virtually at all (if you see what I did there ;) ) will be able to follow along with reasonably well. Fear not! No Discrete Modeling, Statistics, Calculus, or any other high level collegiate mathematics Computer Science majors are forced to endure will be required here. :)

And yet, this is also a book that everyone *needs* to read and understand. Auerbach manages to boil his primary thesis of what meganets are and how they operate into three very simple yet utterly complex words: Volume. Velocity. Virality. And he repeats these words so *very* often that you *will* remember them long after you've read this text. (Though I note this writing this review just 24 hrs after finishing my read of it, and knowing I'll read at least 30 more books before 2022 is done. So check back with me on that after this book actually publishes in about 4.5 months. :D)

Indeed, really the only problem here - potentially corrected before publication - is that at least in the copy I read, the bibliography only accounted for about 15% of the text, which is fairly light for a nonfiction book in my experience, where 20-30% is more normal and 50% is particularly well documented. Thus, the single star deduction.

Still, this truly is a book everyone, from casual readers uninterested in anything computer yet who are forced to use computers in modern life to the uber-techs actually working in and leading the fields in question to the politicians and activists seeking to understand and control these technologies, needs to read. Very much recommended.