If Bernard Hodson has his way, future computer applications will be much smaller than they are today. They will not require the aid of a massive operating system to do their work. And they’ll be more secure than current apps. But judging from history and the modern computer context, Hodson faces an uphill battle in his quest to make technology so efficient.
Hodson is an Ottawa-based theoretical physicist, a ComputerWorld Canada columnist and founder of Genetix Software Inc., a company bent on shaking up the high-tech industry with a new kind of computing, more efficient and secure than the existing paradigm.
Genetix created a computing platform, the Genetix Virtual Machine (VM), different from today’s tech underpinnings, Hodson says.
Right now, stand-alone applications interact with computer resources like memory via an operating system. The OS is something of a traffic cop between apps and resources, ensuring the multiple applications running on a computer have enough memory to complete their tasks.
The Genetix Virtual Machine offers another option: it does away with individual apps and the OS by tying computing tasks closer to the computer itself. The Genetix system presents building blocks from which the user might create apps to perform tasks.
This VM essentially replaces all of the different apps we have now with one app that can do just about anything. “You develop a number of functions with which to build applications,” Hodson explains. “Whenever you want an application, you just see which functions are needed and link them together.”
So far, Hodson has used this “generalized software” approach to run a hospital information system and a 40,000-record database for business credit in Western Canada.
Hodson says the Genetix proposal yields a tiny footprint for tasks. A database normally requiring 200MB of hard drive space would need just 120KB to run in the Genetix VM. “The generated applications are so small that you can put a dozen or 50 or 100 applications into memory at the same time,” Hodson says. “You could then develop a gene to control the flow of the applications.” The upshot: no OS required.
When it comes to computer reliability, security and efficiency, size really does matter. Hodson points out that today’s programs are so bloated with code that errors — bugs — are a foregone conclusion. And thanks to the disconnected nature of computer platforms, it’s relatively easy for hackers and viruses to introduce malicious code into the systems. Not so with the Genetix VM. You can’t add code here.
OSes are notoriously inefficient, Hodson says. “If you look at a typical operating system, only five to 10 per cent of it is ever needed by anybody who ever uses it….What we should be doing is picking the bits and pieces you want. Then you’ll have a nice, compact operating system to work with. Genetix is basically saying the same thing. Pick the pieces and link them.”
The Genetix system takes its cues from a concept way back in high-tech history: the universal machine of Alan Turing. A fellow at King’s College, University of Cambridge in the mid 1930s, Turing described a theoretical computer that operates via a number of symbols, which can be rearranged so the computer performs one task or another. He also described a machine capable of simulating other computers; hence “universal.”
Hodson, who studied under one of Turing’s staff members at the University of Manchester, considered the generalized software approach for some time before he recognized the similarities between his and Turing’s work.
“It was only after a few years that I realized what I was trying to do was develop what Turing had advocated. Then, of course having done that, it became quite interesting to develop it further.” Hodson says 1956 marks the beginning of his generalized software scrutiny, and 1964 is the year his work became genuinely “Turing-like.” The aforementioned credit database came in 1970.
While Hodson has been ferreting out Turing’s ghost (Turing died in 1954), computer technology has taken a different route to the here and now. John von Neumann, a mathematician (1903-1957), devised what’s now known as the von Neumann machine — a computer that separates data from the processing unit and relies on a single storage structure for both the computing instructions and the data.
The von Neumann machine is also known as a stored-program computer.
By treating instructions and data in the same manner, it’s relatively easy to program the machine for different tasks. Unfortunately, it’s also relatively easy to introduce malicious code into the computer, thanks to its re-programmable nature.
Sound familiar? It should. “Virtually every commercially available home computer, microcomputer and supercomputer is a von Neumann machine,” reads the Wikipedia assessment of how entrenched von Neumann’s work is.
“It’s very easy to conceptualize a modern machine working in that framework,” says Allan Borodin, a professor in the University of Toronto’s computer science department.
He explains that von Neumann’s architecture lends itself particularly well to mathematical computations, the sort of work for which computers were primarily used in high-tech’s early days. “The Turing thing’s a little bit harder.”
But that’s not to say the Genetix system is difficult to work with. “Using the rules established for the paradigm, a simple compiler can process any application….In fact the rules are so simple that we can develop an application without the need for a compiler,” Hodson wrote in an article entitled, A New Kind of Computing.
The Genetix VM has its fans. Mark Kuharich, president of MRK International LP, a software development firm, introduced the Genetix concept as “the next big thing” in his e-mail newsletter, The Software View.
But Hodson seems to recognize that not everyone will welcome generalized software with open arms. OS makers and app builders will “fight like fury,” he says, to protect their businesses — profits based on the von Neumann machine and its continued success.
“That’s why I’m suggesting the approach is best through an area where we don’t have this monopolistic group: the smart card and embedded systems market,” Hodson says.
“One of the problems with smart cards right now is the real estate,” he says, explaining why the Genetix VM suits this slice of the IT pie.
“It’s quite small….They have great difficulty running multiple applications.” The Genetix system could solve these problems.
Despite that possibility, however, Hodson has a lot of convincing to do before something like the Genetix VM usurps the existing, von Neumann-esque approach to computing. Nonetheless, he seems optimistic.
“The fact is, something has to be done. The replies I’m getting agree with me so far. They’re all fed up with bloatware, security problems, intrusions. If we can get some people who are interested in the approach, first through smart cards or embedded systems, the probability (of success) would increase quite a bit.”