On Tue, Dec 14, 2004 at 02:39:07AM +0000, Andrew Suffield wrote: > On Mon, Dec 13, 2004 at 02:13:53PM -0600, Kenneth Pronovici wrote: > > On Mon, Dec 13, 2004 at 07:50:02PM +0000, Andrew Suffield wrote: > > > On Mon, Dec 13, 2004 at 11:21:54AM -0800, Bruce Perens wrote: > > > > My surmise is that we'd need an effort like that, raising $250K, to > > > > design and go to full-custom fabrication of an FPLA with fully-open > > > > design. > > > > > > Mine is that one can get useful things done without having to spend > > > ridiculous amounts of money, or even any money at all. Yours is that > > > you can't. Debian proves you wrong every day. > > > > > > There is absolutely no reason why any money is needed for this. Design > > > the damn thing. Somebody will want to produce it. Manufacturing > > > companies would *leap* at the opportunity to make widely desireable > > > chips with zero royalty costs. > > > > I think what you're forgetting (or at least ignoring) is that designing > > hardware is not exactly like designing software. The process is > > similar, yes, but it's not an apples-to-apples comparison. At the > > least, this is because testing your hardware "implementation" is not > > "free" (as in beer). > > Any commercial software company will tell you exactly the same thing > about software: testing is not free. We're *still* here. Consider why > this works (without resorting to things which are obviously not true, > like "current hardware doesn't ship with (many) known bugs", or > "proprietary software is more reliable"). The difference is that software testing is often "free" in a capital sense. I can volunteer my time to test or write open source software, and there is very little capital expense associated with it (my cable modem, my electricity, my PC, etc., much of which has other uses in my household). Testing hardware of this sort requires actually manufacturing it (which is a capital expense) and requires various pieces of test equipment (the purchase of which would also be a capital expense). One way or another, someone will have to bear these expenses. Put it this way: per my hourly rate, I might have "donated" the equivalent of thousands of dollars to Debian this year. However, were I to have to make thousands of dollars of purchases (for fabrication and other speciality hardware) this would be a completely different thing. One is relatively "free" (I have a certain amount of free time in a week) and the other is not (hardware fabrication still costs real money for every iteration no matter how much free time I have). I'm all for advocacy. The problem is that you speak of open source development as some sort of great Truth which has finally become revealed to the Chosen Few who have Seen the Light. I posit that open source development on today's scale did not become practical until PC hardware really became available at "commodity" prices. I do not believe that those sorts prices yet exist in the hardware development world. (Oh, I'm sure you're going to try to slap me with that one. I can't wait. <shiver>) > > I'm not trying to be pessimistic here, just realistic. I think that you > > should be careful not to underestimate these fairly significant > > differences between hardware and software. > > I think you're "underestimating" the difficulty of creating > software. The difference is merely that you happen to be familiar with > a more effective way to do it. Oh, give me a break. > Your point is quite amusing because historically any commercial > organisation would have told you the exact opposite: Yeah, yeah, amusing, right, mm-hmm. > Software is far more expensive to produce than hardware, by several > orders of magnitude. Again, you seem to have missed the difference between capital expenses and labor expenses. KEN -- Kenneth J. Pronovici <pronovic@debian.org>
Attachment:
pgp1agsyZCEhS.pgp
Description: PGP signature