EVERYONE'S AN EXPERT

Dick Pountain/21 January 2001 14:59/Idealog 78

A recent plague of mice in my kitchen set me to thinking, briefly, about the mousetrap. Everyone is familiar with the maxim 'Make a better mousetrap and the world will beat a path to your door' (paraphrased from Ralph Waldo Emerson's longer but more elegant original) and understands it as a hymn to technological progress. However it struck me then that it is patently untrue since no-one appears to have ever invented anything better than a 20p wooden mousetrap the Victorians would recognize. To be sure there are now some non-lethal traps (the size of a dog kennel) but what do you do with the captives? Put them up for adoption? Take them to the Battersea Mouse Home? More likely you tip them over your neighbour's wall and let them beat a path to his door...

For some reason I was reminded of this while reading a white-paper that everyone's talking about by Darek Milhocka (http://www.emulators.com/pentium4.htm) denouncing the Intel Pentium 4 chip for being slower than its predecessors - a flawed mousetrap indeed. Now I have no intention of going into the details here of whether Milhocka's criticisms of the Pentium 4 architecture are sound or not (at a first reading they seem pretty convincing to me). What interested me far more than its details was the tone of Milhocka's paper, which I found familiar and rather disturbing. The tone is passionate ('I urge all computer consumers to BOYCOTT THE PENTIUM 4 and BOYCOTT ALL INTEL PRODUCTS '), bad tempered ('In a related act of stupidity, Intel put 3 integer ALUs in the core') and generally suggests that Milhocka has been personally affronted and injured by this flawed chip design. He continually repeats the mantra that 'clock speed is not everything', since his argument is mostly about cache efficiency, but it's quite clear reading between the lines that for him execution (rather than clock) speed *is* everything - that he, we, have a God-given right to faster and faster processors and that the bastards have tried to rob him of it. The tone is familiar because it is the tone of the Internet Flame Warrior, the tone of those emails I sometimes get from Disgruntled of Tasmania about something I said in this column about Linux, or Windows, or whatever. It is the tone of someone just slightly detached from what I consider to be reality, the reality that I don't actually give a stuff whether I get a faster CPU this month, or next month, or even next year. If I were addicted to playing 3D games I might perhaps feel a mild urge for a few extra Gigaflops, but I don't,  and so the Pentium III in my laptop is already about 10 times overspecified for what I do with it: the difference between a file opening in half a second and one twentieth of a second is neither here nor there to me, and in any case is more probably disk rather than CPU limited.

But something else beyond the personally-aggrieved tone of the Pentium 4 paper fascinates me and that is, how does he know this stuff? Is he an ex-Intel engineer? The site gives no clue, save that he is the founder and CEO of Emulators Inc. and an 80x86 assembler programmer. I know a bit about how difficult this stuff is because for many years I wrote up new processor architectures for Byte - in fact I wrote their first major Pentium overview back in 1993 - and we used to have to read reams, or at very least least quires, of datasheets and do telephone interviews with Intel's architecture engineers. In those days most people were baffled by this stuff and needed articles like that to inform them. Now not only do thousands of people understand this stuff, but they gather together in newsgroups and Web sites to second-guess Intel's engineers. Everyone's an expert, everyone could design a better chip than Intel, despite Intel's spending 3 billion dollars on research in 1999. I'm not denying any of it is true, just stunned that it's possible.

There are of course historical precedents for this, and the most obvious one is the automotive hot-rodding scene that started in the 1950s in the USA. Some people became dissatisfied with the way that Detroit cars looked and went, and so they invented a new hobby which consisted of chopping the roofs off and welding them lower, of reaming out the cylinders and fitting bigger carbs, and generally making them look cool and go faster. Within a decade or so this hobby became an industry in its own right and it became clear that many of these amateur engine tuners new as much or more about those engines than the Detroit engineers who designed them. It strikes me that the PC business has now entered this phase of its evolution. Much of the talk on Cix nowadays is about building your own PCs from motherboards and loose CPU and RAM chips, about which are the fastest combinations and what are the pitfalls. Computers being the abstract machines they are, the cosmetic aspects are at first glance less important: no-one actually chops a tower case and airbrushes burning skeletons on it (that I know of). On the other hand, inside the PC, in the software we now have the concept of 'skins', and I notice that there are 20,866 of them for download on the WinAmp site...