What a Pentagon Report from the Year 2000 Got Right About Cyber War

Ivan Cholakov/Shutterstock.com

Many of the issues that concern cyber strategists today were already clear at the turn of the century.

The defense secretary for a newly elected president is entering his third month in office when a chilling report crosses his desk warning of the catastrophic damage an enemy could visit on the U.S. with a cyberattack.

Such an attack, the report warns, could cripple the U.S. economy. It could strike with no warning. It could be launched asymmetrically by an enemy that’s much weaker than the U.S. in traditional military might.

Even worse, that enemy could use the internet to shield its identity, making it difficult or impossible for U.S. forces to retaliate or to deter an attack before it happens by threatening retaliation inside cyberspace or outside of it.

The defense secretary fires off an email to his pick to lead the Defense Science Board, a civilian adviser to the Pentagon on science and technology matters: “Please take a look at this article, ‘The U.S. is not Safe in a CyberWar,’ and tell me what you think I ought to do about it,” the email asks.

The defense secretary was Donald Rumsfeld. The email was sent in April 2001.

The report, which was released publicly for the first time this month, was oddly prescient about many of the concerns facing cyber strategists today. It also suggests a few alternate paths the U.S. government might have followed if it had responded more forcefully to the growing cyber threat.

“A lot of people told us we were being too alarmist at the time,” Stuart Staniford, the lead author of the 2000 report, told Nextgov.

Nearly 18 years later, Staniford said, he feels “more vindicated than not” about the report’s concerns and conclusions.

When the report came out, it was the height of the first dot-com bubble and companies were rushing to connect as many services and as much information as possible to the internet. What the report tried to point out “is that there are a lot of curveballs in rushing ahead into doing this,” Staniford said.

“But, by and large, people went ahead and did it,” he continued, “and, by and large, bad things continued to happen.”

The Cyber Snowflake

The report, which Staniford co-authored with military and intelligence veterans Sami Saydjari and Ken Williams, was released this month within the first tranche of Rumsfeld’s “snowflakes”—the Pentagon nickname for the short email memos the secretary routinely blasted out to staff and advisers.

Roughly 59,000 pages of such snowflakes and their connected documents are being released in response to a Freedom of Information Act request filed by George Washington University’s National Security Archive.

At the time of the report, Staniford was president of his own company called Silicon Defense, which did contract research with the Defense Advanced Research Projects Agency. Though agency officials approved his request to write the report, Staniford said, they seemed largely uninterested in focusing on what was then a mostly nascent threat.

“I got the feeling that the push wasn’t welcome in the bureaucracy because bureaucracies don’t like people to stir things up,” he said. His emailed response upon first learning about the report’s release was: “OMG I had no idea it went to Rumsfeld.”

Among other things, the report argues that:

  • The basic functions of critical infrastructure, such as dams, energy plants and airports, should be segregated from the internet. This is now considered conventional wisdom, though vulnerable, internet-connected front office tools frequently worm their way into the industrial control systems that operate critical infrastructure.
  • Companies should cooperate with outside digital security researchers, known as white hat hackers, searching for vulnerabilities in their software. This is standard practice for most tech companies now and the Pentagon has run numerous high profile “bug bounty” pilots offering cash payments to white hats who spot hackable vulnerabilities in their systems. Many companies, however, remain highly wary of volunteer bug reports and white hats are frequently threatened with legal action. Bug bounties have yet to take off in the civilian government.
  • Companies should be required to report computer security penetrations. Companies are currently required to report breaches that compromise customers' personal information, but the laws that govern these disclosures vary from state to state. Bills that would create a uniform national breach notification requirement have failed to pass in several congresses. Companies have much more leeway when it comes to reporting breaches that don’t affect customer information.  
  • Over-classification of digital security information by the government might make it difficult to share vital information with companies that they can use to defend themselves. This remains a major problem, especially as the Homeland Security Department tries to ramp up cyber information sharing with the private sector.

The report is cautious about drawing conclusions too confidently.

“We are in a position similar to that of thinkers in 1912 interested in what an air war would be like,” the report notes. “The airplane had been invented, but not yet used in war.”

From the perspective of 2018, Staniford said, things seem mildly clearer than they did in 2000. He cautioned, however, that the U.S. is likely somewhere in the middle of a 40 or 50-year development cycle in cyber warfare and cyber vulnerabilities than at the end of a 20-year process.

Looking Backward

Several things have turned out differently than the report envisioned, said Staniford, who, since 2000, spent half a decade as chief scientist at the cyber threat tracker FireEye and now runs a cyber consultancy.

The specific lines of attack outlined in the report—a computer worm that infects millions of U.S. computers and a distributed denial-of-service attack that knocks a large number of U.S. companies offline—have not yet occurred at the scale the report envisioned.

The report also floats several possible measures that seem short-sighted in retrospect given the scale that both the internet and its vulnerabilities have reached.

The problem of attributing cyber actors, which the report discusses at length, remains a serious problem, but not quite to the extent Staniford envisioned, he said.

The U.S. has attributed some cyber strikes with confidence, such as the 2014 North Korean attack on Sony Pictures Entertainment and the 2016 Russian breach of the Democratic National Committee. That’s mostly because of the National Security Agency's digital surveillance of U.S. adversaries.  

Because the U.S. government can’t share the evidence underpinning those attributions without revealing sensitive intelligence sources and methods, however, those attacking nations are able to plausibly plead innocence on the world stage.

Finally, an economy-crippling attack on the scale that Staniford and his co-authors envisioned remains possible, he said, but due to improvements in cyber defense, it would likely require a lot more money.

It would also require the work of thousands of skilled hackers rather than the few dozen people the report envisions, he said, putting such an attack out of reach for terrorist groups and other non-state actors.

“But thousands of people is not a level of people that a nation-state couldn’t put together, and that’s something people should worry about more than they do,” he said.

Security Still an Afterthought

One thing that has not changed since 2000 is that security is still an afterthought for most software makers, especially smaller ones that are just breaking into the industry.

When Staniford wrote his report in 2000, small companies without security expertise were rushing to take advantage of the dot-com boom. In 2018, companies are similarly rushing into the connected device space—and are securing those devices poorly or not at all.

“You’ve got a new wave of vendors who all go out and hire 20-something engineers who know very little about security,” he said.

And while companies, critical infrastructure providers and the U.S. government are far better secured against digital attacks now than they were in 2000, the country is overall more vulnerable because so much more of our vital information and services are online.

That means that if we’re hit with an attack on the scale that Staniford’s report envisioned, it would be far more damaging now.

“We can see the seeds of where we’ll eventually have to be, but our protections still aren’t remotely adequate,” he said. “Whenever we finally get into a really bad fight with someone in cyberspace, these things will have to grow a lot stronger. But, as to just when that will happen, I don’t have any insight.”