When I asked Eric Raymond, the leader of the Open Source Initiative, what the next big thing is after Linux, he surprised me. He didn't say that I suck. Rarely are enthusiasts able to imagine that there could be a next big thing.
Once upon a time, batch-processing mainframes were the next big thing. They were followed, although not quickly enough, by interactive time-sharing minicomputers, which were followed by stand-alone PCs, which were followed by client/server computing on LANs, which has now been followed by the Internet.
OK, try asking someone what the next big thing is after the Internet. I started to tackle this hard question by answering another similar one about the Ethernet LAN. Ethernet was invented in 1973, and it is, apparently, still going strong. I have to work really hard to answer what the next big thing is after Ethernet.
Answering is complicated by the fact that what we call Ethernet today is not the technology we invented, standardised, and sold for all those years. The original Ethernet was coaxial cable, but now most Ethernets run on twisted pairs and optical fibres. Our Ethernet was of the bus topology, while today Ethernets are stars.
The first Ethernet at Xerox ran at 2.94Mbps and later as an IEEE standard at 10Mbps over shared media, but today it's 100Mbps, going on 1Gbps, over switched media. Our Ethernet patent was controlling packet retransmissions during periods of traffic overload. Today's Ethernets experience collisions only on alternate Palm Sundays.
What the word Ethernet has come to mean is a way of doing business. It's been practiced quite successfully since 1980 and now involves thousands of competitive networking companies.
Our MO with Ethernet is to make a de jure networking standard, then work together to be sure that the open standard results in products that interoperate. Then compete like hell to ship early, be reliable, and drive prices down. Then evolve the standard rapidly according to what's learned in the market. Then build in backward compatibility with the installed base of old products. Then call the new standard Ethernet something.
Ethernet is like Linux in that it is open. It is different from Linux in that users of Ethernet are not invited to mod-ify it willy-nilly. This is the crux of why, despite all the squealing from the Linux mosh pit, Ethernet wins and Linux loses.
So now, if you want to shut down a meeting of Internet enthusiasts, ask them what the next big thing is after IP. Unless they think you're asking about the version after the current IP, Version 4 (IPv4), which will be IPv6, they'll seize up. IP, like Ethernet, was invented in 1973, but it's going to last forever, right?
Well, try this. Will IP survive deployment of the all-optical Internet? I've just read the most persuasive arguments yet for why IP will at least be marginalised if not killed in the foreseeable future.
George Gilder's Technology Report is a $300 per year newsletter (www. gildertech.com). In the July issue, Gilder surveys progress in optical transmission and switching. He projects accelerating progress in wave division multiplexing to fibres carrying 10Gbps on each of 500 wavelengths (lambdas) on each of 140 fibres per sheath or close to petabits (10 to the 15th power bits) per second.
So it's good that somebody is thinking about what the next big thing is after Linux, after Ethernet, and after IP. And no, please, I don't think it's Windows 2000.
Bob Metcalfe invented Ethernet in 1973 and founded 3Com in 1979. e-mail him at firstname.lastname@example.org or visit http://www.idg.net/metcalfe