Last year we read Michael Pollan’s book The Botany of Desire, which traces the history of human relationships with various plants. One of them was the potato. The lowly spud, from its origins in the Andes, had a transformative effect when brought to Europe, because it can produce more calories per acre than any field grain. Its adoption as a staple in Ireland permitted a dramatic expansion of the island’s population.
Of course, it was also responsible for the mass starvation of Irish peasants during the Potato Famine.
Pollan pointed out, however, that at the root of the Potato Famine was the problem of monoculture. Because Europeans had selected and bred the “perfect” potato, the plant didn’t have the great genetic diversity it had possessed in its native land. Instead the whole of Ireland was planted with more-or-less genetically identical exemplars of a single variety of potato. So when the blight struck, all those plants were susceptible — and crops failed everywhere.
It was a signal lesson in the value of biodiversity.
Biological Viruses… and Computer Viruses
What was true of that potato monoculture is also true for global computer networks.
Computer software is even more susceptible to the problem of monoculture because its reproduction is more precise than that of any biological organism. Sexual reproduction ensures that there are slight variances between organisms, even of the same species and breed or variety.
But every installation of Internet Explorer for Windows is identical, almost to the last bit. It is this basic fact that permits hackers to write virus code that can cascade through millions of machines.
The problem is exacerbated by the dominance of Windows, which as an operating system itself provides a monoculture within which hackers can quickly spread disease. Other operating systems — such as Apple's (AAPL) Mac OS, and various distributions of the open-source Linux system — are more intrinsically robust and resistant to viruses.
But none are immune (though a well-implemented Linux installation can come close, which is why it remains very popular for servers). However, the recent Heartbleed debacle (in which a security flaw was revealed in the net’s community-patrolled software for encrypting information from sensitive e-commerce and financial websites) showed that open-source is also capable of dramatic failures. It was a big disappointment for open-source champions.
Since the benefit of networked communication has proven so great, methods have been sought to counter hackers’ handiwork among typical end-users — virus protection software. But the hackers have always seemed to be a step ahead, and gaining ground. The computer security firm Symantec (SYMC) estimates that antivirus software now protects against only about 45 percent of attacks, and recently said that the old model of virus interdiction needed to be radically updated.
We recently read about a novel approach to the problem that may be the beginning of such an update. Michael Franz of the University of California at Irvine proposes to borrow a page from sexual reproduction, so that individual installations of a given piece of software will contain small, random variations — and although they will have no perceptible effect on the program’s performance, they could render it essentially immune to viruses.
A compiler automatically translates applications from the languages that programmers use into the machine code spoken by processors. Franz’ innovation is to have the compiler do its job sub-optimally — resulting in a marginally less efficient implementation of the program, but different enough to render powerless any virus designed to attack that program.
The hackers might adapt, but we surely welcome Dr Franz’ innovation — and can’t wait for it to reach the consumer.
Read Next
