Boundaries

Four months ago an hardware-related article appeared on Stan's blog. It's problematic to create a permanent link, because of the peculiar way which Stan organizes his blog in. If I get it right, this will become the working link in 2008. Why am I citing an article on hardware instead of reviewing a piece of software? Because there will be no boundaries tomorrow. You need to read the whole article, but here is an appetizer:
The development of digital electronics is nowadays almost identical to software development: circuits are designed by geek's sitting at PC workstations, pre-tested using simulator programs, uploaded into re-programmable hardware, and debugged using virtual instruments (scopes and logic analyzers) embedded into the same chip. During the whole process, the engineer never gets up from his/her chair, much less uses a solder iron. Laboratories crammed with carpentry, wires, flat cables, connectors, adaptors, power supplies, soldering stations, and old-fashioned scopes are to a large extent a matter of the past.
The described situation already applies to other fields, but not to NMR. The reason why NMR is so slow to evolve is very simple: we don't buy a new spectrometer every week. If we are rich enough, we change it after 10 years. If we are not so lucky, we can keep it forever! The market is restricted to very few makers, and there's even the risk it becomes a monopoly. The picture described by Stan (and there is no doubt it will materialize in the next decade) is quite different: many labs will be able to assemble a spectrometer by themselves. The assertion that a whole spectrometer can be substituted by a piece of software is an exaggeration. The magnet, the probe, the (pre-)amplifiers cannot be removed (at least in high-resolution NMR), and they account for more than the half of the final price of a spectrometer. They can however be recovered from old instruments. Disregarding this case, if we are free to order the single components from different vendors, instead of being tied to a single manufacturer, their price is going to drop.
The cost-reduction, though important, is not the most exciting thing about SoCs (Systems on Chip). You really have to read that article...