I admire leaders like Yuqing Gao in “Universal Translation” and Daphne Koller in “Bayesian Machine Learning” for their thinking (“10 Emerging Technologies That Will Change Your World,” TR February 2004). I am pleased to see computer scientists contributing to major advances in linguistics and genetics. Perhaps Gao’s team at IBM could benefit from Daphne Koller’s insights and techniques. Could Gao use Koller’s Bayesian tools to systematically analyze the myriads of existing translated texts, in several languages at once, to enrich the context and semantic rules as well as translation patterns in her language database? I am looking forward to a time when I can talk to my computer, in any language.
M. I. Motyl-Szary
I found two of the 10 spotlighted technologies unsettling. “Bayesian Machine Learning” seems a little too much like artificial intelligence for my taste. A search engine that can take data and make inferences about it and then offer recommendations is a scary start to limiting the necessity of humans. And RNAi therapy would make bioweapons even more deadly. Imagine changing someone’s cells so that they won’t have blood clots or so the cells don’t bring oxygen to the brain as they should.
Voting without Paper
Your Q&A with David Dill (“Valid Voting,” TR February 2004), in which he expresses concern about direct-recording electronic voting machines, is timely. Fairfax County, VA, purchased 1,000 of these machines and used them last November. I was an election officer and can attest that the experience was a disaster. In addition to poor human-interface design, many of the Windows-based machines froze up and needed to be rebooted-during voting. Equipping each machine with a printer would increase paper and machine costs tremendously, reward the vendor for a defective, unverifiable design, reduce system reliability, and lead to a paper-processing nightmare during a recount. Simple precinct-based optical scanners are looking better and better.
Throw Out Your Wallet
What do you grab on your way out the door? Michael Schrage, in his article on “Virally Interactive Pixels” (TR February 2004), ignored an opportunity to integrate product functionality. Use the key-chain memory device to hold camera phone data and to interface to a PC, but take it off the key chain and glue it to the phone. Take the car-lock fob off the key chain, glue that to the phone as well, and make it programmable through the phone keypad. Let’s also make the house key a phone-mounted infrared remote and then throw the key chain away. Put credit card and ID info into the phone and then throw the wallet away. In years to come, humans will carry around exactly one electrified information accessory.
I was pleased to see tr raising environmental concerns around nanotechnology, but I was less pleased to read Rodney Brooks’s column on the arrival of robotics (“The Robots Are Here,” TR February 2004). Brooks champions the salvation from migrant workers that robotics can bring, but he does not address the issues raised for those displaced workers. He also advocates robotics for solving the problem of caring for elderly people, but I shudder at the thought of relying on machines as caregivers during my declining years. Technologists and entrepreneurs need to take a global view when promoting new technology. Even the technologies that promise medical benefits have social and environmental costs.
North Reading, MA
Improving Automotive Diagnostics
As someone who does professional fleet maintenance, I read with great interest and sympathy Simson Garfinkel’s column “Deciphering Cars” (TR February 2004). Getting diagnostic trouble codes is relatively easy, but procuring the service bulletins that help a technician track down problems is a challenge. Most manufacturers closely guard these bulletins, so consumers have little chance of obtaining one and will be told to bring the vehicle in for service instead. Any change that yields easier, low-cost diagnostics outside of the dealer shop can’t come along soon enough.
River Falls, WI