Archive users’ personal information securely and indestructibly in a
distributed database spread across the Internet.PlanetLab has built and linked 175 smart nodes at 79 sites in 13 countries,
with a goal of 1,000 nodes by 2006. It is based on Sun Microsystems’ slogan “The
Network [not the box on your desktop] Is the Computer.”
Today’s Internet uses dumb routers to route data packets to the addresses in
the packets’ headers, no questions asked. They can’t tell if a packet is part of
a web page, an email, a file, or a virus. PlanetLab attaches PCs to the routers,
to create a “node” capable of running custom software uploaded by users, which
could perform such analysis and route, or not route, packets accordingly.
We wrote about PlanetLab in
the August issue of HFD (but we misreported its name as “PlanetNet.”) Wade
Roush’s article in Technology Review provides a very detailed explanation
of how it works and its potential.
Reference: Roush, Wade (2003). “The
Internet Reborn.” Technology Review, October.
Quantum Computing Closer
The “qubits” (quantum bits, analogous to the 1s and 0s of conventional
computing bit) required for quantum computing can theoretically be made from a
single subatomic particle, such as an electron. But controlling a single
electron is extremely difficult. US and Swiss researchers have proposed a qubit
made from a group of an odd number of electrons, which can be controlled as one
unit.
Other researchers have proposed qubits that contain three constantly
interacting electrons rather than a pair of electrons and an electrode. When the
energetic state of the middle electron matches the others, the other two can
exchange energy to produce an “on” or “1” signal.
The time frame for quantum computers is generally held to be between ten and
20 years from now.
References: Unknown (2003). “Advance Electron
Teams Make Bigger Qubits.” Technology Research News, September 16; Unknown
(2003). “Quantum Computer
Keeps It Simple.” Technology Research News, August 14.
Exotic Computing and the “Doctor in a cell”
In February, Israeli scientists created three trillion self-contained
computing devices out of DNA molecules, all in a single microliter of salt
solution. Had that tiny drop been programmable, it would have performed 66
gigaflops per second. Scientists are already thinking of harnessing this
capacity to create a “doctor in a cell” that could be injected into a patient,
diagnose any disease present by sensing and analyzing the biological data all
around it, and synthesize and deliver the appropriate drug molecules on the
spot.
Bio-computing is in its infancy. One enzyme-powered DNA computer called MAYA
plays an infallible game of tic-tac-toe, as long as it makes the first move. Not
much compared to a Pentium 4 computer, but it is not so long ago that the infant
Pentium 4 was a mere calculator that could only add, subtract, multiply, and
divide numbers. MAYA could eventually be used to control nanodevices. The method
differs from most DNA computing research efforts, which are aimed at tapping
DNA’s massively parallel nature to solve very large problems. One of those
efforts successfully solved the Traveling Salesman problem.
That success, along with the very idea of DNA computing, belongs to a
University of Southern California computer scientist who recognized that human
cells and computers process and store information in much the same way — in one
case, strings of voltage differentials labeled 0 and 1; in the other, strings of
molecules labeled A, T, C and G. Much hard math and molecular biology later,
test tubes of DNA-laden water called “machines” and “devices” by their creators
are crunching algorithms and producing data.
These tiny computers could be the brains of the “Doctor in a Cell,” and
because genetic material can self-replicate, they could each grow into a
massively parallel supercomputer far more powerful than silicon-based machines
— as they already do, in the form of human beings. Thus, in seeking to
understand and control DNA computing, the researchers are doing no less than
seeking to understand and control life itself.
Less ambitiously, NASA is funding a biology-based machine that will keep
astronauts healthy on deep space missions.
References: Easen, Nick (2003). “Is
life the key to new tech?” CNN, September 19; Unknown (2003). “DNA Plays
Tic-Tac-Toe.” Technology Research News, August 20; Associated Press (2003).
“DNA
may be basis for power computing.” USA Today, August 18.
Wafer-scale Computing
Another approach to refining today’s technology to achieve a significant
advance is Sun Microsystems’ new technology, developed under a defense
supercomputing contract, of chips that communicate wirelessly with one another,
instead of through the relatively thick gold or aluminum wires soldered on
circuit boards. Eliminating interconnecting wires would accelerate inter-chip
data transfers by 60 to 100 times, and enable much more processing power to be
installed per unit of space. The traditional circuit board would become
obsolete.
The Intel Pentium 4 processor can transmit data at about 50 billion bits a
second. Sun says its chip, which is still more of a concept than a product, will
achieve speeds in excess of a trillion bits a second via built-in transmitters
and receivers only a few microns wide that exploit the “capacitive coupling”
effect to send and receive high speed electrical pulses.
Some significant challenges remain to be overcome, but when they are,
computing power will take a major step up.
Reference: Markoff, John (2003). “New Sun
Microsystems Chip May Unseat the Circuit Board.” New York Times, September
22.
International Grid
Chicago’s famed FermiLab, Argonne National Laboratory, and the University of
Chicago are part of an international data grid project that will harness most of
the world’s number-crunching capacity along with enormous amounts of
information, and make it publicly accessible. “It’s a democratization of
science,” said a FermiLab scientist. “If someone wanted to run through a
simulation for whatever purpose, betting on horses, playing the stock market or
doing high-energy physics in Calcutta, they would get access to those services
instead of having to go down and buy their own supercomputer,” said another.
Today, the Web is basically just a massive data warehouse. The grid project
harnesses the massive computing power of the myriad machines that store the
data. Formally, its goal is “to accelerate the handling of the dramatic increase
in the amount of data scientists have to deal with,” according to an Argonne
National Laboratory scientist.
Trial runs between FermiLab and other facilities in the United States and
Europe have already been conducted.
Associated Press (2003). “Scientists
Plan New Supercomputer System.” SiliconValley.com, September 2.
See also the
March issue of HFD for a brief description and discussion of grid
computing.
Another Grid Computing Initiative
A consortium of climatologists from the UK Meteorological Office, industry,
and academia have launched a SETI@Home-type grid computing initiative
that will turn the idle time of perhaps two million PCs around the world into a
supercomputer to predict climate change over the next 50 years. The model will
simulate a period of decades of climate change at a time.
Reference: Best, Jo (2003). “Grid computing used to
predict the future: Over two million users get ready to talk about the
weather…” Silicon.com, September 12.
64-bit PCs Imminent
While awaiting realization of the potential of quantum, molecular, and other
exotic forms of computing, there remain some substantial gains to be made from
conventional technologies. One of those is to bring the power of 64-bit
computing, long available at the mainframe level and not uncommon in small
servers, to the desktop, and that is now happening with the introduction of
Advanced Micro Devices (AMD)’s Athlon 64 processor. Existing 64-bit
processors from Intel (the Itanium) and AMD (the Opteron) are
installed in some servers, but not in end-user PCs. The Athlon 64 marks
the beginning of pervasive 64-bit desktop computing, and although common desktop
applications written to take advantage of the extra power of 64-bit chips won’t
exist until next year, gamers are likely to adopt it quickly.
Reference: Abreu, Elinor Mills (2003). “AMD
Unveils Athlon 64, Seen as Head Start Vs Intel.” Reuters, September 22.
Diamond to Replace Silicon
The condition of heat and pressure under which carbon crystallizes into
diamond are being replicated in Russian-designed machines now turning out
3-carat roughs 24 hours a day, seven days a week; diamond so good that one
Belgian gemologist who examined the stones said it will put DeBeers out of
business. A second company, in Boston, has perfected a completely different
process — chemical vapor deposition — for making near-flawless diamonds and
plans to begin marketing them by year’s end.
Joshua Davis’ article in Wired is well worth the read just for general
interest. Our particular interest, though, is that these processes make possible
the development of inexpensive diamond semiconductors. Not only can diamond run
much hotter than silicon, but also diamond makes a perfect substrate for DNA chips.
Reference: Davis, Joshua (2003). “The New Diamond
Age.” Wired, Issue 11.09, September.
Thinking Machines
Simantha, the simulated surgery patient (see article in the
Practice section of this issue), and other artificial beings will one day
benefit from work conducted at the Sandia National Laboratories to build
computers that can accurately infer intent, remember prior experiences with
users, and allow users to call upon simulated experts to help them analyze
problems and make decisions.
The work began five years ago with DARPA funding and is fundamentally
intended for national security. The original intent — apparently not abandoned
— was to create software replicas of the minds of specific political leaders or
entire populations, which could then be presented with hypothetical situations,
to see how they might react.
For now, the focus is on creating computer models that can help people by
acting more like them. The model will become “an aide tasked with watching
everything you do, learning everything [it] could about you and helping you in
whatever way [it] could.” Knowing what its user is thinking (because it has been
given the same experiences as its user and taught to think the same way) lets a
machine augment the user’s mental abilities by detecting discrepancies between
what the machine is thinking about a given situation and what the user is
thinking. If there is a cognitive dissonance between user and machine, “a
discrepancy alert may be signaled.”
The researchers apparently believe the technology will become ubiquitous and
“allow almost anyone to quickly configure and execute relatively complex
computer simulations.” Not to replace, but to augment, the human mind. The team
has “focused on replicating the processes whereby an individual applies their
unique knowledge to interpret ongoing situations or events. This is a pattern
recognition process that involves episodic memory and emotional processes but
not much of what one would typically consider logical operations.”
Based on their progress over the past five years on methodologies that allow
the knowledge of a specific expert to be captured in computer models, and that
endow synthetic humans with episodic memory (memory of experiences), the Sandia
folks think that cognitive machine technology will be embedded in “most computer
systems” within the next ten years. Since the technology is a tool of national
security, that can only mean “most computer systems associated with US national
security.”
Reference: Delio, Michelle (2003). “Machine Thinks,
Therefore It Is.” Wired News, August 27.
See also previous articles in HFD on: LifeLog,
another DARPA project aimed at capturing everything about a person’s daily life;
collaborative
agents; and AI bots
that help mothers cope with the stresses of caring for cancer-stricken
children.The “Storage of Your Life” and “Spyglasses” articles in the
Devices section of this issue are also relevant.
Machine-to-Machine Communications
M2M (aka machine-to-machine communication, machine networking, pervasive
computing, and control networks) is, like grid computing, a paradigm-shifting
revolutionary technology, once it achieves a critical mass. A swimming pool
chemicals supplier uses M2M to monitor the water quality of swimming pools and
keep a central computer constantly updated via cell phone. It not only provides
an accessible, permanent record of every pool’s temperature and cleanliness, but
also automatically alerts maintenance staff if something needs changing. A
fast-food restaurant chain, which already uses an M2M system to keep track of
how long it takes to complete a customer’s order, plans to use it also to
monitor temperatures inside refrigerators and freezers, to track electricity
use, and determine when it’s time to replace aging equipment. “How do I know
when to replace an air conditioning system because it’s so old it’s costing
money? With this network, I can make those comparisons,” said a company
executive.
Wireless utility meter reading is the biggest application so far. Many
analysts predict rapid growth in the next few years. One survey of some two
dozen M2M firms found they had 23 million devices linked by wireless networks.
Forrester Group predicts that by 2005 there may be as many machines
communicating on wireless networks as people. Intel is aiming to put radios on
all chips. The move is intended to be a disruptive technology that will remake
many existing business models, said a senior Intel executive.
Household appliances are definitely on the radar screen, ready to contact the
maintenance firm with a description of itself and its problem, so the repairman
knows just what to bring and what to do.
Reference: Van, Jon, (2003). “Machine-to-machine
talk not stuff of fiction.” Chicago Tribune, September 2.
End of Animal Experiments and Clinical Trials in
Sight
A life sciences executive predicts that in ten years grid computing will
provide enough power at low enough cost to enable drugs to be modeled in
silico against genomic data. The result will be the end of “all animal
studies,” and of phases I and II clinical studies. Grid computing makes it
unnecessary for institutions to buy supercomputers they might use only
occasionally. Instead, they simply draw as much power as they need from the
grid, over broadband Internet lines.
The growing computing power results in an avalanche of data — a terabyte of
textual information is now being published daily in medical, clinical and
scientific journals — with which traditional structured databases cannot cope.
Unstructured information search tools under development will solve that problem.
Reference: Gifford, Adam (2003). “Computer
modeling future of medicine.” New Zealand Herald, September 3.
See also article in the August issue of
HFD, on in
silico emulations of biological organisms; in the March issue on in
silico surgery on haptic holograms of a patient that may eliminate the
need for experimental in utero surgeries on real mothers and fetuses and
lower the human and financial costs of experimentation; in the August issue on
the data
glut, which is partly the result of increasing use of in silico
experimentation, where binary 1s and 0s simulate the biochemical compounds of
traditional in-vitro petri-dish research; and in the June issue, on unstructured
information searching.
