Dana Blankenhorn

Moore’s Third Law of Management

What other analysts call “big tech” I refer to as “Cloud Czars.”

I do it for a reason.

What distinguishes Apple, Microsoft, Amazon, Google, and Facebook isn’t their managements’ “ruthlessness.” It’s their heavy investment, over many years, in cloud computing centers.

Even early in the 2010s, it took $1 billion of capital each quarter to get into the cloud market. That has only increased with time. Now it averages $4 billion each quarter.

The money must come from somewhere. Cloud is a capital-intensive business. If you’re handing out dividends, it’s because you don’t know what else to do with the shareholders’ cash.

The 2020s: The Machine Internet

(This is Part II of a three-part series on technology in the 2020s, for publication while I’m at the Heidelberg Laureate Forum in Germany.)

The 2010s were the decade of clouds and devices.

But it was done through human intermediaries. People did the input and the output came from people.

Even if you’re talking to your computer, you’re still in there directing it. Our interfaces in 2019 are, in that way, not that different from what they were in 1979. Fingers are replacing mice, voices are replacing keystrokes, but you’re still making the decisions.

That’s what I think will change most in the 2020s.

It all seems frightfully late to me. I wrote about this stuff as The World of Always On as far back as 2002. Then again, I wrote A Guide to Field Computing in 1993, before the iPhone was a gleam in Steve Jobs’ eye.

The Machine Internet has been delayed by the human element. Right now, most “analysts” still assume this will keep it from happening. They’re pushed out the date for self-driving cars from 2019 to 2029. They think autonomous technology must be perfect to be useful.

It only needs to be better than you.

Moore’s Third Law

Almost 20 years ago I began the most important work of my career.

My working title was "Moore’s Lore,” stories about Moore’s Law. A "marketing expert" insisted I put my name in the title. 

The original Moore’s Law, described by Intel co-founder Gordon Moore for Electronics magazine in 1965,  estimated that circuit density could increase for 10 years, at the same cost, and indeed by 1975 Intel was shipping chips that had over 64,000 circuits on them. Moore then “formalized” the law, saying he expected density to continue doubling, at the same pace, for as far as he could foresee. 

That’s Moore’s First Law. Moore never mentioned a second law, but it’s there.

Moore’s Second Law is that just as circuit complexity grows exponentially, so does its capital cost. While you can make up this cost by making more chips, it means the chip-making business is consolidating. Today there are only four manufacturers of microprocessors – Samsung, China’s Taiwan Semiconductor, the Middle East’s Global Foundries, and Intel itself.

The implications of Moore’s Second Law, as it continues grinding down, is that chip production becomes a national security issue.

Ever since Moore’s 1965 article came out, cynics have been trying to call a halt to Moore’s Law. As circuit lines get close together their distance can’t be measured, the cynics note. Every increased expense or shortfall in decreasing circuit size is seen to be evidence that the “Law” is “ending.”

The Era of Systems

Every decade since I began writing has delivered a new era in technology.

The 1970s was the start of the PC era. The 1980s saw the local networking era. The 1990s saw the rise of the Internet. The 2000s was the decade of mobility. This decade has been that of the cloud.

The 2020s will be the era of systems.

We have been talking around this for years. We talk about the Internet of Things. We talk about artificial intelligence, adding it to the cloud to create FOG computing. We talk about blockchain.

We talk about these technologies as silos, but that’s not the way technology works. As I have been writing since publishing Moore’s Lore  in 2001, the exponential improvements driven by chips go into many directions. They pile atop one another, they reinforce one another like electric waves.

PCs combined chips with a user interface. Local Area Networks combined PCs with techniques derived from telephony. The Internet takes these LAN techniques global. Mobility again combined chips with new user interfaces, eliminating peripherals. The cloud era is built on open source software, standardized chips, and the same networking used in the Internet.

The Creative Destruction of Moore’s Law Trumps the Age of Trump.

There has always been an intimate connection between market cycles and political cycles in American politics.

The 2008 crash brought in Barack Obama, and the age we’re now living in. But like most people (including me), he thought the crash and his election meant the end of the crisis, not its beginning.

We’ve all learned better now.

One reason is that the 2008 collapse, and its quick resolution, didn’t drive home the lesson about speculation, and how man-made bubbles can cause enormous hurt when they pop, as they always do. The “Helicopter Ben” Bernanke solution of dropping money on the banks, replacing the losses, then regulating the banks’ use of the funds, only took the banks out of the equation during the next bubble.

The cycle of crisis, recovery, anti-thesis, and excess accelerated, as change does in the age of Moore’s Law. By 2016 the markets were ready for a new period of excess.

Intel Commits to the World of Always On

I began writing about what is now called the Internet of Things, or IoT, over a decade ago, at Corante, a fine site with no business model. It was during that period in the early 2000s where I made zero dollars from writing, but it's a period I'm now quite proud of. 

I called this technology Always On, because it works in the background and doesn't have to be turned on-and-off like a PC.  I wrote about experiments using motes, sensors, and tiny processors affixed to products, or just strewn around the ground in order to study wildlife. Inventory solutions can find your keys with this technology, I wrote, and monitoring solutions can water your lawn.   Health solutions can prevent heart attacks.

Many of these stories focused on Intel, which has now decided that the Internet of Things represents the future. 

Over the last few months Intel began showing retail solutions that can measure customers for clothes at a glance and track clothes in stores. A computer-on-a-chip called Curie is being shown at fashion shows, used for health monitoring and CGI displays that take place around clothes. It’s the flash end of the market. 

But Intel estimates that IoT already represents 4% of the company’s turnover.  The big money is in industry, in turning an Inventory of Things into an Inventory of Systems that can control refineries, collecting big data used to improve production and prevent accidents. It you have ever wondered why there have been so few big planes going down over the U.S. lately, part of the answer lies in this technology, which constantly monitors a plane’s condition and alerts airlines to maintenance needs before conditions grow critical. IDC Research now sees this as a $1.7 trillion market within five years. 

Standards are going to be vital in the IoT, to enable interoperability, and Intel is busy creating and negotiating such standards with software companies as well as carriers. 

A lot of this market turns out to be in Asia, a market Intel failed to crack in its mobility quest. Asian companies can use IoT to help make factories run better, or to monitor both traffic and citizens.Security and privacy implications have many in the U.S. cool to this potential. 

None of this has yet to hit Intel’s stock, which is down almost 17% over the last year as PC sales have slowed and the company’s efforts in mobile have continued to flounder. But Intel still brings 25% of its roughly $15 billion in sales to the net income line each quarter, its cash flow averages over $15 billion per year, and the balance sheet shows over $100 billion in assets with long-term debt that could be instantly repaid, in a crunch, using cash and short-term instruments in hand. Intel, in short, can be here for the long haul.

It’s not often that an investor gets a chance to get in on the “ground floor” of something while at the same time buy a 3.6% yield that can be covered by net income three times. It’s not often you can buy something with a Price/Earnings multiple below that of the market, a safe investment with speculation of leadership in a market niche that is about to explode, as in trillion-dollar explode.

But that is what Intel is looking at, today.

Things with intelligence are becoming parts of intelligent systems, and Intel is ahead of the pack facing this opportunity. Come for the dividend, stay for the speculation.

China Winning the Internet of Systems

Our capitalist system has a way of developing and defining standards that was effective in the 20^th century but may be ineffective early in the 21^st.

We wait for the market to create a need, and to create a leader. Then everyone else gangs up on the leader and creates what they call a standard.

Think of Betamax and VHS. Sony was the first mover, and the other manufacturers ganged up on it to take the consumer market. Or think of Apple and Microsoft in PC operating systems, where Microsoft, IBM, and their OEMs convinced most of the market to wait six full years for a working user interface Apple had introduced in 1984. The same thing is happening with Apple and Google in phones, although in this case Apple has managed to sponge off most of the profits and leave the OEMs toting Google’s laundry and holding the bag.

But there are areas of technology where that process doesn’t work, where only the government can create and enforce standards because the players are too selfish and diffuse to accomplish it themselves.

It’s why the Internet is the way it is. I was covering the story at the time it happened, so I know.

Invisible Technology

What we call “technology” is usually just the icing on a very large cake.

The vast majority of code is invisible. We don’t see it. We don’t know how it works. We don’t know what it does. But it’s absolutely necessary or the whole edifice collapses.

For most of the last generation, however, we’ve obsessed over very visible technology, and created vast fortunes around it. PCs, consumer electronics, and iPhones are all very visible. The last mile of our Internet connections are fairly visible. Applications we use every day, like the word processor I’m using now, are visible. Games are very visible. The Web sites we visit to shop or interact, to learn or to play in, are all very, very visible.

Behind and below that, however, is an enormous amount of invisible technology. Operating systems are mainly invisible. It is when they become visible, when they cause us trouble, that they cause a problem. The Internet Protocol that runs the Internet is invisible. Only those who learn a lot and have mad skilz can understand it.

Last Roadblock to A New World is Removed

Think of this as Volume 16, Number 23 of A-Clue.com, the online newsletter I've written since 1997. Enjoy.

---

In writing about Ipv6 Day, which was Wednesday, reporters struggled to offer a sense of what it means.

We're about to run out of Internet addresses, they said. (Well, there's a workaround, but...) Remaining Ipv4 addresses are rising in price, and we're going to run out of new ones in 2014. (True, but...) Ipv6 is better – more efficient, more secure, inevitable. (As to the last, well, sorta...)

The fact, we've been kicking the can on this for a decade. Ipv6 was actually proposed in 1996, the idea being to expand the address space from four sets of three hexadecimal numbers each – xxx.xxx.xxx.xxx – to eight sets of four – xxxx.xxxx.xxxx.xxxx.xxxx.xxxx.xxxx.xxxx. Instead of having 32 bits – 4.3 billion addresses – you have 128 bits – 3.4 x 10 to the 38^th power addresses. You convert from one to the other by putting zeroes in front of the Ipv4 number.

The Many Markets of Open Source

Even people inside the open source market tend to underestimate it.

They think of it the way they think of the software market. If you're not collecting cash tribute for support (the equivalent of a cash price for the code) you somehow don't count.

By that reckoning open source has been a disappointment. Many companies have difficulty converting downloads into cash. Products on the "bottom of the stack" -- operating systems, databases, languages etc. -- have special problems in the era of "the cloud," with data centers consolidating.

But that's, frankly, a pretty stupid way of looking at it.

• What are Facebook and all the other social media sites but scaled, customer-facing open source applications? They're all written with it.
• Clouds are highly dependent on open source for their very being. Virtualization began with companies wanting to run Windows instances under server Linux. Most of the important cloud creation tools are open source.
• Android is a Linux. It's open source, even if the newest versions may be released first only to partners. Android is beating the iPhone the way Windows beat the Mac.

We got another example in today's news, the launch by Texas Instruments of a site called OpenLink, an open source community for wireless hardware development.