Mr. Gilder is a marvelous writer--I've read much he has written. (My essay on Knowledge & Power is not my best work, but here it is.) I'm deeply sympathetic to Mr. Gilder's argument because he predicts a much more libertarian internet.
Mr. Gilder offers a larger philosophical argument beyond simply business and technology. I enjoyed that discussion, but I restrict my comments here to economic issues.
Famously, Google gives most of its content away for free, or (in comments Gilder credits to Tim Cook) if it's free, you're not the customer; you're the product. That's the least of it. Spanish has two words for "free"--gratis and libre. In our context it means gratis.
Let's count the ways gratis benefits Google:- They are completely immune from any antitrust prosecution and most other regulatory oversight.
- They can roll out buggy, beta software to consumers and improve it over time.
- They don't have to take responsibility for security. Unlike a bank, Google is at no risk if somehow your data gets corrupted or stolen.
- They provide no customer support.
- Your data doesn't belong to you. Instead it belongs to Google, which can monetize it with the help of AI.
- You get locked into a Google world, where everything you own is now at their mercy. (I'm in that situation.) Your data is precisely not libre.
There are some disadvantages.
- It's not really free, but instead of paying with money you pay with time. Attention is the basic currency of Google-world.
- People hate ads. "[O]nly 0.06 percent of smartphone ads were clicked through. Since more than 50 percent of the clicks were by mistake, according to surveys, the intentional response rate was 0.03 percent." This works only for spammers. Ad-blockers are becoming universal.
- Google thinks it can circumvent that by using AI to generate ads that will interest the user. No matter--people still hate them.The result is the value of advertising is declining. Gilder does not believe that AI will ever solve this problem. (I agree with him.)
- Most important--Google loses any information about how valuable its products are. Airlines, for example, respond sensitively to price signals when determining which routes to fly, what equipment to use, what service levels to provide, etc. Price is the best communication mechanism known for conveying economic information. You immediately know what is valuable to consumers, and what isn't.
Google loses all that information by going gratis.
Is Gmail more valuable than Waze? Google has no idea. As a result it has no way of knowing where to invest its money and resources. It's just blindly throwing money at a dartboard.
The above disadvantages are manageable as long as the marginal cost per user is very small. Google invested billions of dollars in huge data farms (Gilder visited the one in The Dalles, OR), located near sources of cheap electricity and cold water (for cooling). Iceland is a prime location for these things. Each center contains thousands of racks of high-end servers connected by millions of miles of fiber-optic cable. This investment has paid off--the marginal cost per search is essentially zero.
But the world is changing fast--according to Gilder the tipping point is the transition from 4G to 5G; from the internet of web pages to the internet of things; from the internet of text and images, to the internet of virtual reality and video rendering; from the internet of computers to the internet of mobile phones, smart watches, and medical devices. To accommodate this new world Google will need to make a massive new investment.
The marginal cost--far from approaching zero--is now becoming near infinite. Without any price information the company has no clue who its most valuable "customers" really are, and cannot determine how best to allocate its resources. Further, the new world opens the door for much nimbler and low-cost competitors. So while Google may be a successful search engine, it will lose its hold over our data. We will own our own data.
Among those nimbler competitors is a Warsaw-based start-up called Golem. Realizing that most CPU clock cycles around the world go unused, the company is trying to tap into that wasted resource. For example, even while I'm busily typing away, my computer's CPU is mostly sitting idle doing nothing. If it were possible for me to rent out those excess clock cycles, then I'd make a little extra cash, and somebody else gets extra computing power they can use.
That's Golem's business model--by installing their software they enable me to rent out my clock cycles in exchange for payment. (In Michael Munger's language Golem is selling a reduction in transaction costs--i.e., making it easier to match my computer resources with somebody who is willing to pay for them.)
Two things need to happen before Golem's business model works. First, my computer needs to be on a 5G network. 5G eliminates the need for fiber-optic cables (at least over the last mile) by enabling high-bandwidth, wireless communication between my computer and the world. Without 5G I can't compete with Google. With 5G, my connection is just as efficient, or--put another way--Google has a few million miles of useless fiber-optic cables on its hands.
Second, Golem needs to handle micropayments--I will be paid pennies per minute of CPU time. The user--who is likely doing a massively parallel calculation--will owe pennies to thousands of computer-owners like me. Golem has got to keep track. The way they do that is through a blockchain currency known as GNT (Golem Network Tokens). The user will buy GNT's from Golem, and use them to pay me and all the others who contribute their clock cycles. As GNT is built on top of bitcoin, at the end of the day I can convert my GNTs to bitcoin, and hence to whatever currency I prefer.
This is more profound than it sounds. In our cloud-centered Google-world, processing and storage have become centralized, as required by fiber-optic technology. But as wireless displaces fiber, then there is no longer a reason for Google's huge data farms--everything can be decentralized. But this new technology requires a new operating system--a different computer architecture. The organizing principle of this new architecture is the blockchain--in all its forms, from distributed ledgers to secure financial transactions.
The blockchain will keep track of who owns what, who is owed what, and how people get paid. A hash of that whole transaction will be embedded in the software as a watermark. My few cents worth of CPU time will be embedded in the output for the ages.
There are currently two competing platforms on top of which the blockchain edifice can be built: bitcoin and ethereum. Bitcoin is very simple and secure--and slow. Ethereum comes with a Turing-complete programming language, in principle making it much easier to use. But it is manifestly less secure.
It turns out Mr. Gilder is agnostic on this choice. That surprises me--from Knowledge & Power I would have expected him to favor the simpler, more reliable bitcoin. Golem, after starting with Ethereum, switched to bitcoin.
I highly recommend reading Life After Google.
Further Reading:
No comments:
Post a Comment