KARP@SUMEX-AIM.STANFORD.EDU (Peter Karp) (10/06/87)
As I expected, a few people disagreed with me on this topic. The following are their comments and my clarification of my initial remarks. Peter A 212) 30-Sep software market... (5453 chars) 214) 30-Sep To: KARP@SUMEX- FSF (4288 chars) FA 221) 1-Oct mlinar%poisson. Re: Free Software Foundation (3239 chars) 225) 2-Oct To: mlinar%pois Re: Free Software Foundation (4568 chars) R FA 231) 3-Oct Eric M. Berg Re: Free Software Foundation software (1814 chars) Message 212 -- ************************ Return-Path: <ASIDONHO%JHUIGF.BITNET@forsythe.stanford.edu> Date: Wed, 30 Sep 87 14:15 EDT From: <ASIDONHO%JHUIGF.BITNET@forsythe.stanford.edu> (Sorry, forgot... lost my memory) Subject: software market... To: karp@sumex-aim.stanford.edu Hi, Thanks for your comments on info-vax. You may get a lot of grief for this. As I read your message, I have to say, I felt like giving you grief as well, but I appreciate your position. It is true that software, free or not, costs money (people cannot, in general, work for free). However, it is more difficult to quantify the value of software than hardware. Perhaps the end user can put a value on the work a particular piece of software does, but I don't think anyone else can, and that includes the COMPANY MARKETING THAT SOFTWARE. I, for one, am sick and tired of overpriced software with overly restrictive licensing policies. The price is too high in several ways: 1) pricing software on CPU type is UNFAIR. Because someone has paid the money to purchase a faster CPU model, does not entitle the vendor to take more from the purchaser. The only fair way I can see to price software is NUMBER OF USERS. We have purchased a new VAX model and have had to pay A LOT more for our software. The machine was purchased to increase the speed of completion of OUR APPLICATIONS primarily, yet every software vendor thinks they deserve a cut! Because we have purchased a machine that can support 100 users, they think we have 100 users paying! They think we have 100 users simultaneously using their software! There are many things some users would like to do that are currently unfeasible, not for technical reasons, but because those users can't pay the amount expected from 100 users. This policy of software is clearly based on the "deep pockets" theory (they bought a big computer so they must have a lot of money to spend), which, loosely translated, is "what can we bilk them for?". 2) the software is just too damn expensive, regardless of CPU type. The technical complexity of the computer market, along with the user fear that natually accompanies such complexity, is a perfect place for these vendors to get away with whatever they can. Many times the software is INCREDIBLY overpriced for what it is, yet sells because it is the only such product (or the only one aggressively marketed). Vendors can claim anything they want, not mention innumerable "gotchas" the user must deal with afterwards, and the industry pundits and "analysts", most of whom know no more about computers than "pie chart", will trumpet this new "revolution" to the four corners of the earth. 3) additional charges, like software maintenance (often required as the software may break with a new OS version), media and documentation ($300 or more for a tape?), long distance phone calls to get the required "password" to run the software (what about when the software breaks because field service swapped a board and the CPU ID register changed? I would like to refuse to buy any software which checks the SID, but often I am pressured into it anyway, hey, I'm not the boss here). The basic problem is the computer software market is always changing, and not well defined in terms of supply and demand. People look at this and try to turn it to their advantage and make a quick buck. The preponderance of overpriced software causes the price of some really good software to go up as well (if they're charging that much for THAT, ours should be worth at least THIS) The way to fix this problem is competition. That is why I love it when I see programs like TEX, EMACS, CMU tcp/ip, PMDF and other really quality stuff for super cheap. Although one could argue that you can't compare these programs (which the authors have chosen to freely distribute) with the product of a commercial company (which they have to sell to survive), this should make those vendors realise they can't get away with it (both the money and the HASSLE). If software like this gets more well known, it must certainly affect the whole policy of the software vendors. The users, by forgoing the expensive, restrictive solution, will be sending these vendors a message that this is no longer a sellers market. The arguments that people are copying software (especially FSF, since they are writing their own stuff, as far as I can see) don't cut it with me. Every human endeavor builds on what came before, and you can't say that because a mathemetician is using calculus means he is copying Newton. Anyway, I am not against commercial software, just sick of being burned, and would like to see the other side (the vendors) get the hot-foot for a change. I would feel a lot better if there was some sort of standard license policy that was fair. Thanks for listening, feel free to include my comments in your summary for the list, as long as I am not completely quoted out of context. -Tom O'Toole Interactive Graphics Facility J.H.U. Medical School Balto. Md. 21205 asidonhopo@jhuigf.bitnet Have you got a B17-6? I'm a bit of a stickler for paperwork... Message 214 -- ************************ Date: Wed, 30 Sep 87 15:40:31 PDT From: Peter Karp <KARP@SUMEX-AIM.STANFORD.EDU> Subject: FSF To: KARP@SUMEX-AIM.STANFORD.EDU > It is true that software, free or not, costs money (people cannot, in > general, work for free). However, it is more difficult to quantify the > value of software than hardware. Actually, I think this question is of equal difficulty for software and hardware. It's easy for the seller to figure out how much it costs to produce and maintain both hardware and software. Both are expensive, for quality software. It is difficult for the seller to figure out how much both hardware and software are worth to a customer. But this is true for just about anything that anyone sells, and the free market is supposed to determine what a fair price is. > 1) pricing software on CPU type is UNFAIR. I agree, number of users is probably a fairer way to price software. We considered this option when preparing a pricing structure for some software I wrote (and I supported it), but it was shot down by arguments to the effect that DEC doesn't provide very fine resolution in terms of number of users for its VMS licenses. Also, how do you measure number of users: total number of people who log in per year? total number of connect or cpu hours they accumulate? total number of hours they use your product? The latter is probably most accurate, but would be a bitch to measure. CPU type is a simple measure which in most cases probably reflects all the above measures. It sounds like your shop is a special case where CPU type is a misleading measure. No measure will be perfect. > This policy of software is clearly based on the "deep pockets" theory No, I don't think so, it's just based on the theory that the more use someone makes of your software, the more they should pay. In practice however this use can be hard to measure. > 2) the software is just too damn expensive, regardless of CPU type. I quite agree with you. And as you say, the solution is competition. However, I would disagree that products like CMU TCP/IP and PMDF are competing. There's a difference between deciding that you will throw away your time and release a piece of software with no support for free, and deciding that you will develope a better, supported piece of software which includes a more reasonable (lower) profit. I would be much happier if CMU supported its TCP/IP and charged a couple thousand dollars for it. > If software like this [free] gets more well known, it must certainly > affect the whole policy of the software vendors. Yeah, the vendors will decide that a bunch of hackers have removed all incentive for the vendors to develop and support better software. You can't compete with a hacker who has no respect for his own abilities and hence gives away his own work for free. Boy, I feel like Ronald Reagan when I say this, but isn't incentive for innovation an important cornerstone of our capitalist society? > The arguments that people are copying software (especially FSF, since > they are writing their own stuff, as far as I can see) don't cut it with > me. Every human endeavor builds on what came before, and you can't say > that because a mathemetician is using calculus means he is copying Newton. It can be a fuzzy line in places but I believe it has to be drawn somewhere. This is what copyrights and patents are for: to reward people for their innovations. > I would feel a lot better if there was some sort of standard license > policy that was fair. Me too. Got any realistic suggestions? Peter ------- Message 221 -- ************************ Return-Path: <mlinar%poisson.usc.edu@oberon.USC.EDU> Date: Thu, 1 Oct 87 18:02:23 PDT From: mlinar%poisson.usc.edu@oberon.USC.EDU (Mitch Mlinar) Message-Id: <8710020102.AA05242@poisson.usc.edu> Subject: Re: Free Software Foundation I disagree. What is wrong with rewriting programs, particularly the buggy ones? I have found that a later version or attempt by another author is often faster, more compact, and has fewer bugs. This keeps the original company on their toes; rather than admit (or deny) any bugs or problems exist (unless they are major), they often figure changes are not worth it because the profit margin is narrowed somewhat. "Healthy" competition keeps the software "state-of-the-art" rather than of "vintage" quality. Thus, your comment about a "new" generation of software is met in reality. Face it, Peter, how much would your life improve with some totally different program? Granted, once in a while a new whiz-banger one comes along. However, won't your productivity increase more rapidly by improving the current batch of programs you have? What if a good part of your software run 20% faster overnight? Or becomes 20% more powerful? Your comment about reinventing the wheel is misleading, it is more like evolution of the wheel. Your last comment about stealing a design is invalid. In the case of FSF (correct me if I am wrong), I have seen compilers and such come out, but none of this "look-and-feel" software and associated legal problems such as GEM and others have run across. They are producing more "generic" tools of high importance. I, for one, would welcome a REAL C compiler for the VAX; the DEC version has some problems in the compiler, and an unlimited number in the library. I have stopped using it. DEC also has a very high NIH factor. Anything that did NOT originate out of DEC itself (C, UN*X, etc.) is treated with total contempt. Even if it is supported, it is poorly supported. Compare that to VMS and FORTRAN which are rigorously supported and have few bugs in comparison. I would be surprised if FSF decided to produce a FORTRAN for the VAX. Think about it. Finally, I totally disagree about "stealing years of experience". Technology, hardware or software, moves ahead. One *should* learn from all the mistakes AND successes made in the past. If you force everyone to build a given package from scratch, very little in the way of new features will be seen. With some of the "thinking" work done for you already, time can be spent solving more difficult or obscure problems. That, my friend, is called progress. ------------- Naturally, Peter, these arguments are NOT directed at you personally. I am just trying to "argue" my side of the issue. Interesting subject, thanks for bringing it up. I am eager to hear what others have to say. -Mitch P.S. You can "net" this if you desire. Message 225 -- ************************ Mail-From: KARP created at 2-Oct-87 11:08:50 Date: Fri, 2 Oct 87 11:08:47 PDT From: Peter Karp <KARP@SUMEX-AIM.STANFORD.EDU> Subject: Re: Free Software Foundation To: mlinar%poisson.usc.edu@OBERON.USC.EDU Hi, Well, maybe I need to clarify my position a bit. 1) Regardless of the issues of stealing software that I brought up, I am against the concept of a foundation whose purpose is to provide free software. I see no reason to place a value on hardware but none on software when both are necessary to solve a problem. Producing and maintaining high quality software requires a lot of highly skilled labor, so I see no reason not to charge for it. This does not mean that I feel it is reasonable to overcharge for software, and I have absolutely no qualms with fair competition, i.e., a company's efforts to produce and maintain high quality software at a lower cost than an existing company. In fact, I am involved in such an effort. But as far as I can tell, the FSF is not competing fairly - they are simply giving away their efforts. In addition, when you don't charge for software you have little incentive to support it, and I believe support is an important thing. 2) Regarding the issue of stealing software, I realize this is a tricky one and I hoped to inspire debate. I think there are a lot of fine lines to be drawn here. Let's consider a few cases. My general feeling is that it's alright for a company to duplicate the functionality of a piece of software as long as they don't use proprietary design techniques developed by the original authors. If an element of a piece of software is explicitly in the public domain it's hard to argue against reproducing it. In fact I don't find this definition completely acceptable, but it's the best I can do for now. Some examples: GNU Emacs. I would guess that Stallman's original Tops-20 Emacs is in the public domain, so I see no problem with their producing GNU and VMS versions of it. Anyone else is free to do this as well. Now I am a bit annoyed that they are distributing this for free since I believe they are competing unfairly with what I consider to be another quality Emacs, namely Gosling Emacs, which I believe is being sold for a reasonable price by some company. GNU C compiler. I guess my original message was misleading since it was in response to David Kashtan's message about the GNU C compiler. Since the description of the C language is in the public domain, anyone is free to write a C compiler. And I'd be quite happy to see someone compete with DEC's overpriced and poorly implemented C compiler and runtime system, as long as they don't steal any of DEC's implementation techniques (which I would guess would be difficult since they probably don't release the source). Once again though, I'm annoyed that FSF is giving this away. What would be wrong with their charging say $2000/copy? (A fraction of DEC's current price?) Isn't a C compiler worth this? GNU itself, i.e., new implementation of Unix. This is where I believe the thievery is going on. From what I understand, FSF wants to duplicate all the Unix utilities (including the C-shell?). Neither the specification nor the source of these programs is in the public domain. But both are easily available. FSF will obviously copy the specifications of these programs (i.e., their behavior), and will probably copy the implementation techniques by looking at the code for these programs, even if they don't copy the code line for line. The distinction I'm making between Unix and the C compiler is that generally the *behavior* of C compilers is public domain, i.e., a description of the syntax and semantics of the C language is public knowledge. But the behavior of the Unix utilities is not in the public domain: AT&T put a lot of effort into figuring out what is a good set of utilities to have, and how to implement and interface them conceptually. They deserve to be rewarded for their incentive and effort in this area. Now I'm no lawyer and I don't know what kind of legal ground I'm standing on here, but this seems to me to be analagous to patenting a new process or device. The general point here is that if people are free to copy your hard-won techniques, they rob you of some of the incentive for developing new techniques in the first place. Peter ------- Message 231 -- ************************ Return-Path: <A.ERIC@GSB-WHY.Stanford.EDU> Date: Sat 3 Oct 87 22:33:41-PDT From: Eric M. Berg <A.Eric@GSB-WHY.Stanford.EDU> Subject: Re: Free Software Foundation software To: KARP@SUMEX-AIM.Stanford.EDU Reply-To: A.ERIC@GSB-HOW.STANFORD.EDU One could also argue that they are ripping off the original authors of all the programs they copy. A large amount of the effort in building any large system is designing it at a conceptual level: figuring out what how to decompose it into manageable pieces and how it should interface to the environment within which it resides. FSF is stealing this design when they copy existing software. They are also stealing the sometimes years of experience that have gone into shaping a sophisticated piece of software. Peter: I suspect that it was precisely the feeling of being ripped off that led Stallman to the idea of the Free Software Foundation. After all, look at hom many people are making money off of EMACS (esp. for the Un*x environment, e.g. Unipress & its competitors), the original (PDP-10) version of which was the product of many people, Stallman chief among them. It's not hard to imagine that Stallman might feel that *his* work was being ripped off for commerical purposes, leading him to want to create a better product and give it away free (GnuEmacs). Eric (P.S. -- You're welcome to summarize this in the replies you post to INFO-VAX, altho' I'm sure someone else has already written in about it.) ------- -------
roland@LBL-RTSG.ARPA (Roland McGrath) (10/07/87)
I'm a bit disgusted at the mentality of Mr. Karp-- that everyone should be a ruthless capitolist and proud of it. Hasn't the man ever heard of generosity? I happen to feel damn good that I've written about 1000 lines of C code in Make-- from my head and Stallman's ideas-- for the benefit of the general computer community, not to mention most of an almost complete ANSI standard C library. He wants me to sell all of this just because I probably could?? Not to mention that one thing he said was absolutely positively wrong and very insulting to me and other people contributing their time and ideas to GNU and the FSF-- he said that we would probably look at the Unix source to copy ideas; despite the fact that 95% of the time, this is blantantly illegal, it is also completely contrary to the wishes of Mr. Stallman. Sincerely, Roland McGrath (roland@lbl-rtsg.arpa) [Please direct any responses to the address above, as I don't subscribe to this list or read it on Usenet.]