| 1 | (For more information about the GNU project and free software, |
| 2 | look at the files `GNU', `COPYING', and `DISTRIB', in the same |
| 3 | directory as this file.) |
| 4 | |
| 5 | |
| 6 | Why Software Should Be Free |
| 7 | |
| 8 | by Richard Stallman |
| 9 | |
| 10 | (Version of April 24, 1992) |
| 11 | |
| 12 | Copyright (C) 1991, 1992, Free Software Foundation, Inc. |
| 13 | Verbatim copying and redistribution is permitted |
| 14 | without royalty; alteration is not permitted. |
| 15 | |
| 16 | Introduction |
| 17 | ************ |
| 18 | |
| 19 | The existence of software inevitably raises the question of how |
| 20 | decisions about its use should be made. For example, suppose one |
| 21 | individual who has a copy of a program meets another who would like a |
| 22 | copy. It is possible for them to copy the program; who should decide |
| 23 | whether this is done? The individuals involved? Or another party, |
| 24 | called the "owner"? |
| 25 | |
| 26 | Software developers typically consider these questions on the |
| 27 | assumption that the criterion for the answer is to maximize developers' |
| 28 | profits. The political power of business has led to the government |
| 29 | adoption of both this criterion and the answer proposed by the |
| 30 | developers: that the program has an owner, typically a corporation |
| 31 | associated with its development. |
| 32 | |
| 33 | I would like to consider the same question using a different |
| 34 | criterion: the prosperity and freedom of the public in general. |
| 35 | |
| 36 | This answer cannot be decided by current law--the law should conform |
| 37 | to ethics, not the other way around. Nor does current practice decide |
| 38 | this question, although it may suggest possible answers. The only way |
| 39 | to judge is to see who is helped and who is hurt by recognizing owners |
| 40 | of software, why, and how much. In other words, we should perform a |
| 41 | cost-benefit analysis on behalf of society as a whole, taking account of |
| 42 | individual freedom as well as production of material goods. |
| 43 | |
| 44 | In this essay, I will describe the effects of having owners, and show |
| 45 | that the results are detrimental. My conclusion is that programmers |
| 46 | have the duty to encourage others to share, redistribute, study and |
| 47 | improve the software we write: in other words, to write "free" |
| 48 | software.(1) |
| 49 | |
| 50 | How Owners Justify Their Power |
| 51 | ****************************** |
| 52 | |
| 53 | Those who benefit from the current system where programs are property |
| 54 | offer two arguments in support of their claims to own programs: the |
| 55 | emotional argument and the economic argument. |
| 56 | |
| 57 | The emotional argument goes like this: "I put my sweat, my heart, my |
| 58 | soul into this program. It comes from *me*, it's *mine*!" |
| 59 | |
| 60 | This argument does not require serious refutation. The feeling of |
| 61 | attachment is one that programmers can cultivate when it suits them; it |
| 62 | is not inevitable. Consider, for example, how willingly the same |
| 63 | programmers usually sign over all rights to a large corporation for a |
| 64 | salary; the emotional attachment mysteriously vanishes. By contrast, |
| 65 | consider the great artists and artisans of medieval times, who didn't |
| 66 | even sign their names to their work. To them, the name of the artist |
| 67 | was not important. What mattered was that the work was done--and the |
| 68 | purpose it would serve. This view prevailed for hundreds of years. |
| 69 | |
| 70 | The economic argument goes like this: "I want to get rich (usually |
| 71 | described inaccurately as `making a living'), and if you don't allow me |
| 72 | to get rich by programming, then I won't program. Everyone else is like |
| 73 | me, so nobody will ever program. And then you'll be stuck with no |
| 74 | programs at all!" This threat is usually veiled as friendly advice |
| 75 | from the wise. |
| 76 | |
| 77 | I'll explain later why this threat is a bluff. First I want to |
| 78 | address an implicit assumption that is more visible in another |
| 79 | formulation of the argument. |
| 80 | |
| 81 | This formulation starts by comparing the social utility of a |
| 82 | proprietary program with that of no program, and then concludes that |
| 83 | proprietary software development is, on the whole, beneficial, and |
| 84 | should be encouraged. The fallacy here is in comparing only two |
| 85 | outcomes--proprietary software vs. no software--and assuming there are |
| 86 | no other possibilities. |
| 87 | |
| 88 | Given a system of intellectual property, software development is |
| 89 | usually linked with the existence of an owner who controls the |
| 90 | software's use. As long as this linkage exists, we are often faced |
| 91 | with the choice of proprietary software or none. However, this linkage |
| 92 | is not inherent or inevitable; it is a consequence of the specific |
| 93 | social/legal policy decision that we are questioning: the decision to |
| 94 | have owners. To formulate the choice as between proprietary software |
| 95 | vs. no software is begging the question. |
| 96 | |
| 97 | The Argument against Having Owners |
| 98 | ********************************** |
| 99 | |
| 100 | The question at hand is, "Should development of software be linked |
| 101 | with having owners to restrict the use of it?" |
| 102 | |
| 103 | In order to decide this, we have to judge the effect on society of |
| 104 | each of those two activities *independently*: the effect of developing |
| 105 | the software (regardless of its terms of distribution), and the effect |
| 106 | of restricting its use (assuming the software has been developed). If |
| 107 | one of these activities is helpful and the other is harmful, we would be |
| 108 | better off dropping the linkage and doing only the helpful one. |
| 109 | |
| 110 | To put it another way, if restricting the distribution of a program |
| 111 | already developed is harmful to society overall, then an ethical |
| 112 | software developer will reject the option of doing so. |
| 113 | |
| 114 | To determine the effect of restricting sharing, we need to compare |
| 115 | the value to society of a restricted (i.e., proprietary) program with |
| 116 | that of the same program, available to everyone. This means comparing |
| 117 | two possible worlds. |
| 118 | |
| 119 | This analysis also addresses the simple counterargument sometimes |
| 120 | made that "the benefit to the neighbor of giving him or her a copy of a |
| 121 | program is cancelled by the harm done to the owner." This |
| 122 | counterargument assumes that the harm and the benefit are equal in |
| 123 | magnitude. The analysis involves comparing these magnitudes, and shows |
| 124 | that the benefit is much greater. |
| 125 | |
| 126 | To elucidate this argument, let's apply it in another area: road |
| 127 | construction. |
| 128 | |
| 129 | It would be possible to fund the construction of all roads with |
| 130 | tolls. This would entail having toll booths at all street corners. |
| 131 | Such a system would provide a great incentive to improve roads. It |
| 132 | would also have the virtue of causing the users of any given road to |
| 133 | pay for that road. However, a toll booth is an artificial obstruction |
| 134 | to smooth driving--artificial, because it is not a consequence of how |
| 135 | roads or cars work. |
| 136 | |
| 137 | Comparing free roads and toll roads by their usefulness, we find that |
| 138 | (all else being equal) roads without toll booths are cheaper to |
| 139 | construct, cheaper to run, safer, and more efficient to use.(2) In a |
| 140 | poor country, tolls may make the roads unavailable to many citizens. |
| 141 | The roads without toll booths thus offer more benefit to society at |
| 142 | less cost; they are preferable for society. Therefore, society should |
| 143 | choose to fund roads in another way, not by means of toll booths. Use |
| 144 | of roads, once built, should be free. |
| 145 | |
| 146 | When the advocates of toll booths propose them as *merely* a way of |
| 147 | raising funds, they distort the choice that is available. Toll booths |
| 148 | do raise funds, but they do something else as well: in effect, they |
| 149 | degrade the road. The toll road is not as good as the free road; giving |
| 150 | us more or technically superior roads may not be an improvement if this |
| 151 | means substituting toll roads for free roads. |
| 152 | |
| 153 | Of course, the construction of a free road does cost money, which the |
| 154 | public must somehow pay. However, this does not imply the inevitability |
| 155 | of toll booths. We who must in either case pay will get more value for |
| 156 | our money by buying a free road. |
| 157 | |
| 158 | I am not saying that a toll road is worse than no road at all. That |
| 159 | would be true if the toll were so great that hardly anyone used the |
| 160 | road--but this is an unlikely policy for a toll collector. However, as |
| 161 | long as the toll booths cause significant waste and inconvenience, it is |
| 162 | better to raise the funds in a less obstructive fashion. |
| 163 | |
| 164 | To apply the same argument to software development, I will now show |
| 165 | that having "toll booths" for useful software programs costs society |
| 166 | dearly: it makes the programs more expensive to construct, more |
| 167 | expensive to distribute, and less satisfying and efficient to use. It |
| 168 | will follow that program construction should be encouraged in some other |
| 169 | way. Then I will go on to explain other methods of encouraging and (to |
| 170 | the extent actually necessary) funding software development. |
| 171 | |
| 172 | The Harm Done by Obstructing Software |
| 173 | ===================================== |
| 174 | |
| 175 | Consider for a moment that a program has been developed, and any |
| 176 | necessary payments for its development have been made; now society must |
| 177 | choose either to make it proprietary or allow free sharing and use. |
| 178 | Assume that the existence of the program and its availability is a |
| 179 | desirable thing.(3) |
| 180 | |
| 181 | Restrictions on the distribution and modification of the program |
| 182 | cannot facilitate its use. They can only interfere. So the effect can |
| 183 | only be negative. But how much? And what kind? |
| 184 | |
| 185 | Three different levels of material harm come from such obstruction: |
| 186 | |
| 187 | * Fewer people use the program. |
| 188 | |
| 189 | * None of the users can adapt or fix the program. |
| 190 | |
| 191 | * Other developers cannot learn from the program, or base new work |
| 192 | on it. |
| 193 | |
| 194 | Each level of material harm has a concomitant form of psychosocial |
| 195 | harm. This refers to the effect that people's decisions have on their |
| 196 | subsequent feelings, attitudes and predispositions. These changes in |
| 197 | people's ways of thinking will then have a further effect on their |
| 198 | relationships with their fellow citizens, and can have material |
| 199 | consequences. |
| 200 | |
| 201 | The three levels of material harm waste part of the value that the |
| 202 | program could contribute, but they cannot reduce it to zero. If they |
| 203 | waste nearly all the value of the program, then writing the program |
| 204 | harms society by at most the effort that went into writing the program. |
| 205 | Arguably a program that is profitable to sell must provide some net |
| 206 | direct material benefit. |
| 207 | |
| 208 | However, taking account of the concomitant psychosocial harm, there |
| 209 | is no limit to the harm that proprietary software development can do. |
| 210 | |
| 211 | Obstructing Use of Programs |
| 212 | =========================== |
| 213 | |
| 214 | The first level of harm impedes the simple use of a program. A copy |
| 215 | of a program has nearly zero marginal cost (and you can pay this cost by |
| 216 | doing the work yourself), so in a free market, it would have nearly zero |
| 217 | price. A license fee is a significant disincentive to use the program. |
| 218 | If a widely-useful program is proprietary, far fewer people will use it. |
| 219 | |
| 220 | It is easy to show that the total contribution of a program to |
| 221 | society is reduced by assigning an owner to it. Each potential user of |
| 222 | the program, faced with the need to pay to use it, may choose to pay, |
| 223 | or may forego use of the program. When a user chooses to pay, this is a |
| 224 | zero-sum transfer of wealth between two parties. But each time someone |
| 225 | chooses to forego use of the program, this harms that person without |
| 226 | benefiting anyone. The sum of negative numbers and zeros must be |
| 227 | negative. |
| 228 | |
| 229 | But this does not reduce the amount of work it takes to *develop* |
| 230 | the program. As a result, the efficiency of the whole process, in |
| 231 | delivered user satisfaction per hour of work, is reduced. |
| 232 | |
| 233 | This reflects a crucial difference between copies of programs and |
| 234 | cars, chairs, or sandwiches. There is no copying machine for material |
| 235 | objects outside of science fiction. But programs are easy to copy; |
| 236 | anyone can produce as many copies as are wanted, with very little |
| 237 | effort. This isn't true for material objects because matter is |
| 238 | conserved: each new copy has to be built from raw materials in the same |
| 239 | way that the first copy was built. |
| 240 | |
| 241 | With material objects, a disincentive to use them makes sense, |
| 242 | because fewer objects bought means less raw materials and work needed |
| 243 | to make them. It's true that there is usually also a startup cost, a |
| 244 | development cost, which is spread over the production run. But as long |
| 245 | as the marginal cost of production is significant, adding a share of the |
| 246 | development cost does not make a qualitative difference. And it does |
| 247 | not require restrictions on the freedom of ordinary users. |
| 248 | |
| 249 | However, imposing a price on something that would otherwise be free |
| 250 | is a qualitative change. A centrally-imposed fee for software |
| 251 | distribution becomes a powerful disincentive. |
| 252 | |
| 253 | What's more, central production as now practiced is inefficient even |
| 254 | as a means of delivering copies of software. This system involves |
| 255 | enclosing physical disks or tapes in superfluous packaging, shipping |
| 256 | large numbers of them around the world, and storing them for sale. This |
| 257 | cost is presented as an expense of doing business; in truth, it is part |
| 258 | of the waste caused by having owners. |
| 259 | |
| 260 | Damaging Social Cohesion |
| 261 | ======================== |
| 262 | |
| 263 | Suppose that both you and your neighbor would find it useful to run a |
| 264 | certain program. In ethical concern for your neighbor, you should feel |
| 265 | that proper handling of the situation will enable both of you to use it. |
| 266 | A proposal to permit only one of you to use the program, while |
| 267 | restraining the other, is divisive; neither you nor your neighbor should |
| 268 | find it acceptable. |
| 269 | |
| 270 | Signing a typical software license agreement means betraying your |
| 271 | neighbor: "I promise to deprive my neighbor of this program so that I |
| 272 | can have a copy for myself." People who make such choices feel |
| 273 | internal psychological pressure to justify them, by downgrading the |
| 274 | importance of helping one's neighbors--thus public spirit suffers. |
| 275 | This is psychosocial harm associated with the material harm of |
| 276 | discouraging use of the program. |
| 277 | |
| 278 | Many users unconsciously recognize the wrong of refusing to share, so |
| 279 | they decide to ignore the licenses and laws, and share programs anyway. |
| 280 | But they often feel guilty about doing so. They know that they must |
| 281 | break the laws in order to be good neighbors, but they still consider |
| 282 | the laws authoritative, and they conclude that being a good neighbor |
| 283 | (which they are) is naughty or shameful. That is also a kind of |
| 284 | psychosocial harm, but one can escape it by deciding that these licenses |
| 285 | and laws have no moral force. |
| 286 | |
| 287 | Programmers also suffer psychosocial harm knowing that many users |
| 288 | will not be allowed to use their work. This leads to an attitude of |
| 289 | cynicism or denial. A programmer may describe enthusiastically the |
| 290 | work that he finds technically exciting; then when asked, "Will I be |
| 291 | permitted to use it?", his face falls, and he admits the answer is no. |
| 292 | To avoid feeling discouraged, he either ignores this fact most of the |
| 293 | time or adopts a cynical stance designed to minimize the importance of |
| 294 | it. |
| 295 | |
| 296 | Since the age of Reagan, the greatest scarcity in the United States |
| 297 | is not technical innovation, but rather the willingness to work together |
| 298 | for the public good. It makes no sense to encourage the former at the |
| 299 | expense of the latter. |
| 300 | |
| 301 | Obstructing Custom Adaptation of Programs |
| 302 | ========================================= |
| 303 | |
| 304 | The second level of material harm is the inability to adapt programs. |
| 305 | The ease of modification of software is one of its great advantages over |
| 306 | older technology. But most commercially available software isn't |
| 307 | available for modification, even after you buy it. It's available for |
| 308 | you to take it or leave it, as a black box--that is all. |
| 309 | |
| 310 | A program that you can run consists of a series of numbers whose |
| 311 | meaning is obscure. No one, not even a good programmer, can easily |
| 312 | change the numbers to make the program do something different. |
| 313 | |
| 314 | Programmers normally work with the "source code" for a program, which |
| 315 | is written in a programming language such as Fortran or C. It uses |
| 316 | names to designate the data being used and the parts of the program, and |
| 317 | it represents operations with symbols such as `+' for addition and `-' |
| 318 | for subtraction. It is designed to help programmers read and change |
| 319 | programs. Here is an example; a program to calculate the distance |
| 320 | between two points in a plane: |
| 321 | |
| 322 | float |
| 323 | distance (p0, p1) |
| 324 | struct point p0, p1; |
| 325 | { |
| 326 | float xdist = p1.x - p0.x; |
| 327 | float ydist = p1.y - p0.y; |
| 328 | return sqrt (xdist * xdist + ydist * ydist); |
| 329 | } |
| 330 | |
| 331 | Here is the same program in executable form, on the computer I |
| 332 | normally use: |
| 333 | |
| 334 | 1314258944 -232267772 -231844864 1634862 |
| 335 | 1411907592 -231844736 2159150 1420296208 |
| 336 | -234880989 -234879837 -234879966 -232295424 |
| 337 | 1644167167 -3214848 1090581031 1962942495 |
| 338 | 572518958 -803143692 1314803317 |
| 339 | |
| 340 | Source code is useful (at least potentially) to every user of a |
| 341 | program. But most users are not allowed to have copies of the source |
| 342 | code. Usually the source code for a proprietary program is kept secret |
| 343 | by the owner, lest anybody else learn something from it. Users receive |
| 344 | only the files of incomprehensible numbers that the computer will |
| 345 | execute. This means that only the program's owner can change the |
| 346 | program. |
| 347 | |
| 348 | A friend once told me of working as a programmer in a bank for about |
| 349 | six months, writing a program similar to something that was commercially |
| 350 | available. She believed that if she could have gotten source code for |
| 351 | that commercially available program, it could easily have been adapted |
| 352 | to their needs. The bank was willing to pay for this, but was not |
| 353 | permitted to--the source code was a secret. So she had to do six |
| 354 | months of make-work, work that counts in the GNP but was actually waste. |
| 355 | |
| 356 | The MIT Artificial Intelligence lab (AI lab) received a graphics |
| 357 | printer as a gift from Xerox around 1977. It was run by free software |
| 358 | to which we added many convenient features. For example, the software |
| 359 | would notify a user immediately on completion of a print job. Whenever |
| 360 | the printer had trouble, such as a paper jam or running out of paper, |
| 361 | the software would immediately notify all users who had print jobs |
| 362 | queued. These features facilitated smooth operation. |
| 363 | |
| 364 | Later Xerox gave the AI lab a newer, faster printer, one of the first |
| 365 | laser printers. It was driven by proprietary software that ran in a |
| 366 | separate dedicated computer, so we couldn't add any of our favorite |
| 367 | features. We could arrange to send a notification when a print job was |
| 368 | sent to the dedicated computer, but not when the job was actually |
| 369 | printed (and the delay was usually considerable). There was no way to |
| 370 | find out when the job was actually printed; you could only guess. And |
| 371 | no one was informed when there was a paper jam, so the printer often |
| 372 | went for an hour without being fixed. |
| 373 | |
| 374 | The system programmers at the AI lab were capable of fixing such |
| 375 | problems, probably as capable as the original authors of the program. |
| 376 | Xerox was uninterested in fixing them, and chose to prevent us, so we |
| 377 | were forced to accept the problems. They were never fixed. |
| 378 | |
| 379 | Most good programmers have experienced this frustration. The bank |
| 380 | could afford to solve the problem by writing a new program from |
| 381 | scratch, but a typical user, no matter how skilled, can only give up. |
| 382 | |
| 383 | Giving up causes psychosocial harm--to the spirit of self-reliance. |
| 384 | It is demoralizing to live in a house that you cannot rearrange to suit |
| 385 | your needs. It leads to resignation and discouragement, which can |
| 386 | spread to affect other aspects of one's life. People who feel this way |
| 387 | are unhappy and do not do good work. |
| 388 | |
| 389 | Imagine what it would be like if recipes were hoarded in the same |
| 390 | fashion as software. You might say, "How do I change this recipe to |
| 391 | take out the salt?", and the great chef would respond, "How dare you |
| 392 | insult my recipe, the child of my brain and my palate, by trying to |
| 393 | tamper with it? You don't have the judgment to change my recipe and |
| 394 | make it work right!" |
| 395 | |
| 396 | "But my doctor says I'm not supposed to eat salt! What can I do? |
| 397 | Will you take out the salt for me?" |
| 398 | |
| 399 | "I would be glad to do that; my fee is only $50,000." Since the |
| 400 | owner has a monopoly on changes, the fee tends to be large. "However, |
| 401 | right now I don't have time. I am busy with a commission to design a |
| 402 | new recipe for ship's biscuit for the Navy Department. I might get |
| 403 | around to you in about two years." |
| 404 | |
| 405 | Obstructing Software Development |
| 406 | ================================ |
| 407 | |
| 408 | The third level of material harm affects software development. |
| 409 | Software development used to be an evolutionary process, where a person |
| 410 | would take an existing program and rewrite parts of it for one new |
| 411 | feature, and then another person would rewrite parts to add another |
| 412 | feature; in some cases, this continued over a period of twenty years. |
| 413 | Meanwhile, parts of the program would be "cannibalized" to form the |
| 414 | beginnings of other programs. |
| 415 | |
| 416 | The existence of owners prevents this kind of evolution, making it |
| 417 | necessary to start from scratch when developing a program. It also |
| 418 | prevents new practitioners from studying existing programs to learn |
| 419 | useful techniques or even how large programs can be structured. |
| 420 | |
| 421 | Owners also obstruct education. I have met bright students in |
| 422 | computer science who have never seen the source code of a large |
| 423 | program. They may be good at writing small programs, but they can't |
| 424 | begin to learn the different skills of writing large ones if they can't |
| 425 | see how others have done it. |
| 426 | |
| 427 | In any intellectual field, one can reach greater heights by standing |
| 428 | on the shoulders of others. But that is no longer generally allowed in |
| 429 | the software field--you can only stand on the shoulders of the other |
| 430 | people *in your own company*. |
| 431 | |
| 432 | The associated psychosocial harm affects the spirit of scientific |
| 433 | cooperation, which used to be so strong that scientists would cooperate |
| 434 | even when their countries were at war. In this spirit, Japanese |
| 435 | oceanographers abandoning their lab on an island in the Pacific |
| 436 | carefully preserved their work for the invading U.S. Marines, and left a |
| 437 | note asking them to take good care of it. |
| 438 | |
| 439 | Conflict for profit has destroyed what international conflict spared. |
| 440 | Nowadays scientists in many fields don't publish enough in their papers |
| 441 | to enable others to replicate the experiment. They publish only enough |
| 442 | to let readers marvel at how much they were able to do. This is |
| 443 | certainly true in computer science, where the source code for the |
| 444 | programs reported on is usually secret. |
| 445 | |
| 446 | It Does Not Matter How Sharing Is Restricted |
| 447 | ============================================ |
| 448 | |
| 449 | I have been discussing the effects of preventing people from copying, |
| 450 | changing and building on a program. I have not specified how this |
| 451 | obstruction is carried out, because that doesn't affect the conclusion. |
| 452 | Whether it is done by copy protection, or copyright, or licenses, or |
| 453 | encryption, or ROM cards, or hardware serial numbers, if it *succeeds* |
| 454 | in preventing use, it does harm. |
| 455 | |
| 456 | Users do consider some of these methods more obnoxious than others. |
| 457 | I suggest that the methods most hated are those that accomplish their |
| 458 | objective. |
| 459 | |
| 460 | Software Should be Free |
| 461 | ======================= |
| 462 | |
| 463 | I have shown how ownership of a program--the power to restrict |
| 464 | changing or copying it--is obstructive. Its negative effects are |
| 465 | widespread and important. It follows that society shouldn't have |
| 466 | owners for programs. |
| 467 | |
| 468 | Another way to understand this is that what society needs is free |
| 469 | software, and proprietary software is a poor substitute. Encouraging |
| 470 | the substitute is not a rational way to get what we need. |
| 471 | |
| 472 | Vaclav Havel has advised us to "Work for something because it is |
| 473 | good, not just because it stands a chance to succeed." A business |
| 474 | making proprietary software stands a chance of success in its own narrow |
| 475 | terms, but it is not what is good for society. |
| 476 | |
| 477 | Why People Will Develop Software |
| 478 | ******************************** |
| 479 | |
| 480 | If we eliminate intellectual property as a means of encouraging |
| 481 | people to develop software, at first less software will be developed, |
| 482 | but that software will be more useful. It is not clear whether the |
| 483 | overall delivered user satisfaction will be less; but if it is, or if |
| 484 | we wish to increase it anyway, there are other ways to encourage |
| 485 | development, just as there are ways besides toll booths to raise money |
| 486 | for streets. Before I talk about how that can be done, first I want to |
| 487 | question how much artificial encouragement is truly necessary. |
| 488 | |
| 489 | Programming is Fun |
| 490 | ================== |
| 491 | |
| 492 | There are some lines of work that few will enter except for money; |
| 493 | road construction, for example. There are other fields of study and |
| 494 | art in which there is little chance to become rich, which people enter |
| 495 | for their fascination or their perceived value to society. Examples |
| 496 | include mathematical logic, classical music, and archaeology; and |
| 497 | political organizing among working people. People compete, more sadly |
| 498 | than bitterly, for the few funded positions available, none of which is |
| 499 | funded very well. They may even pay for the chance to work in the |
| 500 | field, if they can afford to. |
| 501 | |
| 502 | Such a field can transform itself overnight if it begins to offer the |
| 503 | possibility of getting rich. When one worker gets rich, others demand |
| 504 | the same opportunity. Soon all may demand large sums of money for doing |
| 505 | what they used to do for pleasure. When another couple of years go by, |
| 506 | everyone connected with the field will deride the idea that work would |
| 507 | be done in the field without large financial returns. They will advise |
| 508 | social planners to ensure that these returns are possible, prescribing |
| 509 | special privileges, powers and monopolies as necessary to do so. |
| 510 | |
| 511 | This change happened in the field of computer programming in the past |
| 512 | decade. Fifteen years ago, there were articles on "computer |
| 513 | addiction": users were "onlining" and had hundred-dollar-a-week habits. |
| 514 | It was generally understood that people frequently loved programming |
| 515 | enough to break up their marriages. Today, it is generally understood |
| 516 | that no one would program except for a high rate of pay. People have |
| 517 | forgotten what they knew fifteen years ago. |
| 518 | |
| 519 | When it is true at a given time that most people will work in a |
| 520 | certain field only for high pay, it need not remain true. The dynamic |
| 521 | of change can run in reverse, if society provides an impetus. If we |
| 522 | take away the possibility of great wealth, then after a while, when the |
| 523 | people have readjusted their attitudes, they will once again be eager |
| 524 | to work in the field for the joy of accomplishment. |
| 525 | |
| 526 | The question, "How can we pay programmers?", becomes an easier |
| 527 | question when we realize that it's not a matter of paying them a |
| 528 | fortune. A mere living is easier to raise. |
| 529 | |
| 530 | Funding Free Software |
| 531 | ===================== |
| 532 | |
| 533 | Institutions that pay programmers do not have to be software houses. |
| 534 | Many other institutions already exist which can do this. |
| 535 | |
| 536 | Hardware manufacturers find it essential to support software |
| 537 | development even if they cannot control the use of the software. In |
| 538 | 1970, much of their software was free because they did not consider |
| 539 | restricting it. Today, their increasing willingness to join |
| 540 | consortiums shows their realization that owning the software is not |
| 541 | what is really important for them. |
| 542 | |
| 543 | Universities conduct many programming projects. Today, they often |
| 544 | sell the results, but in the 1970s, they did not. Is there any doubt |
| 545 | that universities would develop free software if they were not allowed |
| 546 | to sell software? These projects could be supported by the same |
| 547 | government contracts and grants which now support proprietary software |
| 548 | development. |
| 549 | |
| 550 | It is common today for university researchers to get grants to |
| 551 | develop a system, develop it nearly to the point of completion and call |
| 552 | that "finished", and then start companies where they really finish the |
| 553 | project and make it usable. Sometimes they declare the unfinished |
| 554 | version "free"; if they are thoroughly corrupt, they instead get an |
| 555 | exclusive license from the university. This is not a secret; it is |
| 556 | openly admitted by everyone concerned. Yet if the researchers were not |
| 557 | exposed to the temptation to do these things, they would still do their |
| 558 | research. |
| 559 | |
| 560 | Programmers writing free software can make their living by selling |
| 561 | services related to the software. I have been hired to port the GNU C |
| 562 | compiler to new hardware, and to make user-interface extensions to GNU |
| 563 | Emacs. (I offer these improvements to the public once they are done.) |
| 564 | I also teach classes for which I am paid. |
| 565 | |
| 566 | I am not alone in working this way; there is now a successful, |
| 567 | growing corporation which does no other kind of work. Several other |
| 568 | companies also provide commercial support for the free software of the |
| 569 | GNU system. This is the beginning of the independent software support |
| 570 | industry-an industry that could become quite large if free software |
| 571 | becomes prevalent. It provides users with an option generally |
| 572 | unavailable for proprietary software, except to the very wealthy. |
| 573 | |
| 574 | New institutions such as the Free Software Foundation can also fund |
| 575 | programmers. Most of the foundation's funds come from users buying |
| 576 | tapes through the mail. The software on the tapes is free, which means |
| 577 | that every user has the freedom to copy it and change it, but many |
| 578 | nonetheless pay to get copies. (Recall that "free software" refers to |
| 579 | freedom, not to price.) Some users order tapes who already have a copy, |
| 580 | as a way of making a contribution they feel we deserve. The Foundation |
| 581 | also receives sizable donations from computer manufacturers. |
| 582 | |
| 583 | The Free Software Foundation is a charity, and its income is spent on |
| 584 | hiring as many programmers as possible. If it had been set up as a |
| 585 | business, distributing the same free software to the public for the same |
| 586 | fee, it would now provide a very good living for its founder. |
| 587 | |
| 588 | Because the Foundation is a charity, programmers often work for the |
| 589 | Foundation for half of what they could make elsewhere. They do this |
| 590 | because we are free of bureaucracy, and because they feel satisfaction |
| 591 | in knowing that their work will not be obstructed from use. Most of |
| 592 | all, they do it because programming is fun. In addition, volunteers |
| 593 | have written many useful programs for us. (Recently even technical |
| 594 | writers have begun to volunteer.) |
| 595 | |
| 596 | This confirms that programming is among the most fascinating of all |
| 597 | fields, along with music and art. We don't have to fear that no one |
| 598 | will want to program. |
| 599 | |
| 600 | What Do Users Owe to Developers? |
| 601 | ================================ |
| 602 | |
| 603 | There is a good reason for users of software to feel a moral |
| 604 | obligation to contribute to its support. Developers of free software |
| 605 | are contributing to the users' activities, and it is both fair and in |
| 606 | the long term interest of the users to give them funds to continue. |
| 607 | |
| 608 | However, this does not apply to proprietary software developers, |
| 609 | since obstructionism deserves a punishment rather than a reward. |
| 610 | |
| 611 | We thus have a paradox: the developer of useful software is entitled |
| 612 | to the support of the users, but any attempt to turn this moral |
| 613 | obligation into a requirement destroys the basis for the obligation. A |
| 614 | developer can either deserve a reward or demand it, but not both. |
| 615 | |
| 616 | I believe that an ethical developer faced with this paradox must act |
| 617 | so as to deserve the reward, but should also entreat the users for |
| 618 | voluntary donations. Eventually the users will learn to support |
| 619 | developers without coercion, just as they have learned to support public |
| 620 | radio and television stations. |
| 621 | |
| 622 | What Is Software Productivity? |
| 623 | ****************************** |
| 624 | |
| 625 | If software were free, there would still be programmers, but perhaps |
| 626 | fewer of them. Would this be bad for society? |
| 627 | |
| 628 | Not necessarily. Today the advanced nations have fewer farmers than |
| 629 | in 1900, but we do not think this is bad for society, because the few |
| 630 | deliver more food to the consumers than the many used to do. We call |
| 631 | this improved productivity. Free software would require far fewer |
| 632 | programmers to satisfy the demand, because of increased software |
| 633 | productivity at all levels: |
| 634 | |
| 635 | * Wider use of each program that is developed. |
| 636 | |
| 637 | * The ability to adapt existing programs for customization instead |
| 638 | of starting from scratch. |
| 639 | |
| 640 | * Better education of programmers. |
| 641 | |
| 642 | * The elimination of duplicate development effort. |
| 643 | |
| 644 | Those who object to cooperation because it would result in the |
| 645 | employment of fewer programmers, are actually objecting to increased |
| 646 | productivity. Yet these people usually accept the widely-held belief |
| 647 | that the software industry needs increased productivity. How is this? |
| 648 | |
| 649 | "Software productivity" can mean two different things: the overall |
| 650 | productivity of all software development, or the productivity of |
| 651 | individual projects. Overall productivity is what society would like to |
| 652 | improve, and the most straightforward way to do this is to eliminate the |
| 653 | artificial obstacles to cooperation which reduce it. But researchers |
| 654 | who study the field of "software productivity" focus only on the |
| 655 | second, limited, sense of the term, where improvement requires difficult |
| 656 | technological advances. |
| 657 | |
| 658 | Is Competition Inevitable? |
| 659 | ************************** |
| 660 | |
| 661 | Is it inevitable that people will try to compete, to surpass their |
| 662 | rivals in society? Perhaps it is. But competition itself is not |
| 663 | harmful; the harmful thing is *combat*. |
| 664 | |
| 665 | There are many ways to compete. Competition can consist of trying to |
| 666 | achieve ever more, to outdo what others have done. For example, in the |
| 667 | old days, there was competition among programming wizards--competition |
| 668 | for who could make the computer do the most amazing thing, or for who |
| 669 | could make the shortest or fastest program for a given task. This kind |
| 670 | of competition can benefit everyone, *as long as* the spirit of good |
| 671 | sportsmanship is maintained. |
| 672 | |
| 673 | Constructive competition is enough competition to motivate people to |
| 674 | great efforts. A number of people are competing to be the first to have |
| 675 | visited all the countries on Earth; some even spend fortunes trying to |
| 676 | do this. But they do not bribe ship captains to strand their rivals on |
| 677 | desert islands. They are content to let the best person win. |
| 678 | |
| 679 | Competition becomes combat when the competitors begin trying to |
| 680 | impede each other instead of advancing themselves--when "Let the best |
| 681 | person win" gives way to "Let me win, best or not." Proprietary |
| 682 | software is harmful, not because it is a form of competition, but |
| 683 | because it is a form of combat among the citizens of our society. |
| 684 | |
| 685 | Competition in business is not necessarily combat. For example, when |
| 686 | two grocery stores compete, their entire effort is to improve their own |
| 687 | operations, not to sabotage the rival. But this does not demonstrate a |
| 688 | special commitment to business ethics; rather, there is little scope for |
| 689 | combat in this line of business short of physical violence. Not all |
| 690 | areas of business share this characteristic. Withholding information |
| 691 | that could help everyone advance is a form of combat. |
| 692 | |
| 693 | Business ideology does not prepare people to resist the temptation to |
| 694 | combat the competition. Some forms of combat have been made banned with |
| 695 | anti-trust laws, truth in advertising laws, and so on, but rather than |
| 696 | generalizing this to a principled rejection of combat in general, |
| 697 | executives invent other forms of combat which are not specifically |
| 698 | prohibited. Society's resources are squandered on the economic |
| 699 | equivalent of factional civil war. |
| 700 | |
| 701 | "Why Don't You Move to Russia?" |
| 702 | ******************************* |
| 703 | |
| 704 | In the United States, any advocate of other than the most extreme |
| 705 | form of laissez-faire selfishness has often heard this accusation. For |
| 706 | example, it is leveled against the supporters of a national health care |
| 707 | system, such as is found in all the other industrialized nations of the |
| 708 | free world. It is leveled against the advocates of public support for |
| 709 | the arts, also universal in advanced nations. The idea that citizens |
| 710 | have any obligation to the public good is identified in America with |
| 711 | Communism. But how similar are these ideas? |
| 712 | |
| 713 | Communism as was practiced in the Soviet Union was a system of |
| 714 | central control where all activity was regimented, supposedly for the |
| 715 | common good, but actually for the sake of the members of the Communist |
| 716 | party. And where copying equipment was closely guarded to prevent |
| 717 | illegal copying. |
| 718 | |
| 719 | The American system of intellectual property exercises central |
| 720 | control over distribution of a program, and guards copying equipment |
| 721 | with automatic copying protection schemes to prevent illegal copying. |
| 722 | |
| 723 | By contrast, I am working to build a system where people are free to |
| 724 | decide their own actions; in particular, free to help their neighbors, |
| 725 | and free to alter and improve the tools which they use in their daily |
| 726 | lives. A system based on voluntary cooperation, and decentralization. |
| 727 | |
| 728 | Thus, if we are to judge views by their resemblance to Russian |
| 729 | Communism, it is the software owners who are the Communists. |
| 730 | |
| 731 | The Question of Premises |
| 732 | ************************ |
| 733 | |
| 734 | I make the assumption in this paper that a user of software is no |
| 735 | less important than an author, or even an author's employer. In other |
| 736 | words, their interests and needs have equal weight, when we decide |
| 737 | which course of action is best. |
| 738 | |
| 739 | This premise is not universally accepted. Many maintain that an |
| 740 | author's employer is fundamentally more important than anyone else. |
| 741 | They say, for example, that the purpose of having owners of software is |
| 742 | to give the author's employer the advantage he deserves--regardless of |
| 743 | how this may affect the public. |
| 744 | |
| 745 | It is no use trying to prove or disprove these premises. Proof |
| 746 | requires shared premises. So most of what I have to say is addressed |
| 747 | only to those who share the premises I use, or at least are interested |
| 748 | in what their consequences are. For those who believe that the owners |
| 749 | are more important than everyone else, this paper is simply irrelevant. |
| 750 | |
| 751 | But why would a large number of Americans accept a premise which |
| 752 | elevates certain people in importance above everyone else? Partly |
| 753 | because of the belief that this premise is part of the legal traditions |
| 754 | of American society. Some people feel that doubting the premise means |
| 755 | challenging the basis of society. |
| 756 | |
| 757 | It is important for these people to know that this premise is not |
| 758 | part of our legal tradition. It never has been. |
| 759 | |
| 760 | Thus, the Constitution says that the purpose of copyright is to |
| 761 | "promote the progress of science and the useful arts." The Supreme |
| 762 | Court has elaborated on this, stating in `Fox Film vs. Doyal' that "The |
| 763 | sole interest of the United States and the primary object in conferring |
| 764 | the [copyright] monopoly lie in the general benefits derived by the |
| 765 | public from the labors of authors." |
| 766 | |
| 767 | We are not required to agree with the Constitution or the Supreme |
| 768 | Court. (At one time, they both condoned slavery.) So their positions |
| 769 | do not disprove the owner supremacy premise. But I hope that the |
| 770 | awareness that this is a radical right-wing assumption rather than a |
| 771 | traditionally recognized one will weaken its appeal. |
| 772 | |
| 773 | Conclusion |
| 774 | ********** |
| 775 | |
| 776 | We like to think that our society encourages helping your neighbor; |
| 777 | but each time we reward someone for obstructionism, or admire them for |
| 778 | the wealth they have gained in this way, we are sending the opposite |
| 779 | message. |
| 780 | |
| 781 | Software hoarding is one form of our general willingness to disregard |
| 782 | the welfare of society for personal gain. We can trace this disregard |
| 783 | from Ronald Reagan to Jim Bakker, from Ivan Boesky to Exxon, from |
| 784 | failing banks to failing schools. We can measure it with the size of |
| 785 | the homeless population and the prison population. The antisocial |
| 786 | spirit feeds on itself, because the more we see that other people will |
| 787 | not help us, the more it seems futile to help them. Thus society decays |
| 788 | into a jungle. |
| 789 | |
| 790 | If we don't want to live in a jungle, we must change our attitudes. |
| 791 | We must start sending the message that a good citizen is one who |
| 792 | cooperates when appropriate, not one who is successful at taking from |
| 793 | others. I hope that the free software movement will contribute to |
| 794 | this: at least in one area, we will replace the jungle with a more |
| 795 | efficient system which encourages and runs on voluntary cooperation. |
| 796 | |
| 797 | ---------- Footnotes ---------- |
| 798 | |
| 799 | (1) The word "free" in "free software" refers to freedom, not to |
| 800 | price; the price paid for a copy of a free program may be zero, or |
| 801 | small, or (rarely) quite large. |
| 802 | |
| 803 | (2) The issues of pollution and traffic congestion do not alter |
| 804 | this conclusion. If we wish to make driving more expensive to |
| 805 | discourage driving in general, it is disadvantageous to do this using |
| 806 | toll booths, which contribute to both pollution and congestion. A tax |
| 807 | on gasoline is much better. Likewise, a desire to enhance safety by |
| 808 | limiting maximum speed is not relevant; a free access road enhances the |
| 809 | average speed by avoiding stops and delays, for any given speed limit. |
| 810 | |
| 811 | (3) One might regard a particular computer program as a harmful |
| 812 | thing that should not be available at all, like the Lotus Marketplace |
| 813 | database of personal information, which was withdrawn from sale due to |
| 814 | public disapproval. Most of what I say does not apply to this case, |
| 815 | but it makes little sense to argue for having an owner on the grounds |
| 816 | that the owner will make the program less available. The owner will |
| 817 | not make it *completely* unavailable, as one would wish in the case of |
| 818 | a program whose use is considered destructive. |
| 819 | |