I came across a note about Soviet radio amateurs: people reconstructed the circuitry of the British ZX Spectrum from a single chip borrowed for a weekend, probing it with an oscilloscope for months. They reproduced the design—but suing anyone over it was impossible in principle. No one held the rights to it: not the Soviet engineers, nor, for that matter, Sinclair itself within the Soviet legal system.
The next question asked itself: if a program was neither a commodity nor an object of legal rights, how did the state keep track of it at all? It turned out that it did—diligently, under GOST standards, with inventory numbers and test reports. There was an entire system of algorithm and program funds; by the late 1980s, it united around eighty organizations, and within it a program followed a path more like the biography of a machine tool than the biography of a text. I checked the publication archive: neither “GosFAP” nor “the algorithm and program fund” has been examined here to this day.
The first thing you learn from reading the documents is that the system had several birth dates, all of them administrative. State Committee for Science and Technology Resolution No. 28 of February 21, 1966 launched the sectoral funds. Resolution No. 443 of 1969 and No. 258 of June 5, 1975 expanded the structure. And Resolution No. 581 of December 10, 1979 brought it all together into a single system—the State Fund of Algorithms and Programs, or GosFAP—and put the “Regulations on the State Fund of Algorithms and Programs” into effect from March 1, 1980.
Its structure was territorial-sectoral. A given ministry or agency got exactly one sectoral fund; a Union republic got exactly one republican fund, plus intersectoral specialized funds by profile. Above them all stood the Central Information Fund, administered by the All-Union Scientific and Technical Information Center—VNTITs. The lead organization for scientific and methodological guidance was made VNIIPOU under the State Committee for Science and Technology, the very institute that later wrote Instruction RD 50-244-81 on what paperwork had to accompany a program.
The scale by the end of the decade: around 80 funds, including 8 intersectoral, 10 republican, and 62 sectoral ones. In other words, the country’s software had a full-fledged, specified-down-to-the-last-instruction system for storage and distribution—and that was long before it had any kind of legal right attached to it.
One clause of the Regulations stands apart. A program documented under ESPD standards and assigned an inventory number by the Central Fund could be counted as… a publication. The wording is explicit: such algorithms were “equated with published works” under the Regulations on the Awarding of Academic Degrees. An inventory number at VNTITs worked like a registration certificate today—except that what was registered here was the program itself, not its author.
Instruction RD 50-244-81 lists the accompanying documents. A copy of the interagency acceptance-test report. A pilot-operation report. An expert-review report. A GosFAP information card. An implementation report. And in the middle of that list—a statement of the actual development costs.
This is where things get interesting. The developer was required to confirm how much it had cost to write the program—yet it was transferred to the fund free of charge. Clause 25 of the Regulations leaves no room for qualification: “The transfer by developer organizations of software tools to GosFAP shall be carried out free of charge.” Submission to the fund was mandated from above: under GOST 19.102-77, at the implementation stage all tools developed under State Committee for Science and Technology assignments and under plans at various levels were to be transferred, as were those whose transfer was specified in the technical requirements.
There was money in the system after all—but at the other end of the pipe. The user paid for the fund’s services: information support, delivery of a copy on a machine-readable medium, staff training, installing the program on the customer’s machine. Payment followed standard or sectoral rates, or, where those did not exist, cost estimates—and development costs were specifically excluded from those estimates. The state did not pay for writing the program. It paid for rewriting it.
And one more detail that looks very strange in 1979. The fund transferred program copies “without granting the right to reproduce them for other organizations.” The holder of the original, upon requests from the funds, issued duplicates and separately granted the right to make copies for distribution among users. That is exactly what we would call a license today: the medium separately, the right to reproduce separately. A structure the rest of the world invented as a commercial instrument grew up here as an element of accounting.
A logical question follows: did any of this actually get reproduced at scale? By the documents—poorly. The weak point is already visible in the Regulations themselves: the funds maintained “records and analysis of needs” and “calculation and analysis of the economic effect from the use of software tools.” Need was not tested by demand; it was planned. The effect calculation, too, was a planned figure that had to be presented.
By the end of the decade, this was plain to see. Makarevich, writing in the 1988 issue of Software Products and Systems, says it outright: programs submitted to GosFAP were often low-quality, had very limited reproducibility, and did not measure up to a “market” software product, while the chief flaw was outdated organizational and economic mechanisms. A survey of development tools from the same period gives a figure that makes you uncomfortable: of all the tools examined, 105—58 percent—were transferred to various funds, 35 were not transferred at all, and for 43 there is simply no data. On the circulation of tools transferred to the funds, the survey honestly says that no data exists.
Then the system began eating itself. If getting a ready-made tool through a fund costs more and takes longer than writing your own, people write their own. The 1991 review “The State of Software Development Technology in the USSR” says it in plain language: the user is forced to create proprietary, in many cases “homegrown,” tools for their own needs, and this finally reduces the software industry to a secondary role. Fifty-eight percent submitted, zero data on circulation, and a line of customers who went off to write code behind their own partition wall.
The Soviet scheme looks less strange if you remember that the world was debating how to treat programs in roughly the same years and with the same outcome—only in the opposite direction.
In 1955, users of IBM’s most powerful machine at the time pooled their efforts into an organization called SHARE and set up a common program library. Thomas Haigh, tracing its history, lists a set of traits we would now call open source: a library, shared coding and documentation standards, discussion, bug reports to the author, acceptance of improvements back into the common code. All that in the late fifties. In the 1960s, SHARE tried to bolt academic peer review onto the library—and failed; by the early seventies, machine vendors, universities, and commercial firms had pushed it out of the market for quality mathematical procedures.
The turning point has an exact date: June 23, 1969, when IBM announced separate charges for hardware and software, and “free software included with the package” came to an end. From there the process went legal: an amendment to U.S. copyright law appeared in 1980; in England, France, Hungary, and Japan—in the first half of the eighties.
Meanwhile, the USSR went its own way, with the sign reversed. In 1984, State Committee for Science and Technology Resolution No. 41 equated programs with objects of new technology, giving developers the right to bonuses of up to six monthly salaries per year—but treated them as industrial and technical products, not as objects of copyright. It was in this document that the term “software product” officially appeared: the product arrived five years before the right to it. In 1987, the State Committee for Computing Technology and Informatics, GKVTI, was created; in 1988 came Council of Ministers Resolution No. 511 and a package of some thirty regulatory documents, including the “Regulations on the Registration and Protection of the Copyrights of Software Developers.” The package suffered from two congenital defects: it was internal to the bureaucracy, and it treated programs as objects of both copyright and property law at once.
Beginning in 1990, software registration was handled by the All-Union Scientific Production Association “Algorithm” under GKVTI. In November 1991, its functions were transferred to the new Russian Agency for the Legal Protection of Programs—RosAPO; in February 1993, a presidential decree reassigned it to the patent office. An institution that had spent thirty years counting programs as manufactured articles was retrained, in two years, to count them as works.
Russian Federation Law No. 3523-1 “On the Legal Protection of Computer Programs and Databases” was adopted on September 23, 1992. Virtually everything that came before it ended up beyond the line. Supreme Soviet Resolution No. 3524-1 on the procedure for bringing it into force says in Clause Three that the law “applies to relations connected with the creation and use of programs arising after the entry into force of the said Law.”
That is, the entire Soviet software history—millions of lines, whole schools of thought, tens of thousands of items submitted to funds with inventory numbers—remained outside the new legal framework.
And Clause Four of the same resolution is that compromise patch always sewn on in such cases. Until January 1, 1994, programs made public before the law came into force could be used for scientific, educational, and personal purposes without profit, without the author’s consent and without payment of remuneration. Fifteen months of transition window, after which the country was supposed to remember overnight who had written what.
The dates, incidentally, diverge across the sources. A teaching text on the history of legal protection says the law entered into force on October 20, 1992, while Resolution No. 3524-1 itself says “from the day of publication.” I found both versions and do not consider the distinction fundamental, but I am honestly leaving both on the table.
It would seem to be the end: the law appeared, the registry died, full stop. But the registry turned out differently.
In 2008, the idea of recreating a repository for software developments was voiced at the Siberian Branch of the Russian Academy of Sciences, and there the Algorithm and Program Fund of the Siberian Branch of the Russian Academy of Sciences appeared—with product review, registration, a catalog, a corporate repository, and even a “virtual demonstration platform” where you could get your hands on a program without installing it yourself. And in 2013, Russian Government Resolution No. 62 established the National Fund of Algorithms and Programs: a federal system for collecting, processing, and storing programs created or acquired with the involvement of federal budget funds.
The mechanics of this fund are almost a word-for-word retelling of Clause 25 of the 1979 Regulations. Budget-funded agencies must give priority to what is already in the fund; fund objects are made available for free use under a standard licensing agreement; the goal is described in plain words as reducing public authorities’ technological dependence through previously created and published programs. In 2021, when the government approved new regulations for the fund, news reports cited the figure: 303 objects. The fund’s portal today shows 780.
From eighty funds to 780 objects in thirteen years—a kind of progress it is hard to say anything definite about without knowing how many such programs were written over the same years with budget money. I found no data on that. Nor did I find a figure for how many entries GosFAP held at its peak: there is a GosFAP software catalog as of July 1, 1988, published in 1989 and split into two volumes—for ES EVM computers and separately for SM EVM computers, microcomputers, and personal machines—but library descriptions do not disclose the catalog’s print run, and I do not, naturally, have a paper copy. So it remains unknown exactly how many programs passed through that repository. To me, that blank seems more honest than any elegant estimate.
It remains to say where this all began. The same Kommersant, in its piece on Spectrum clones, offers a detail I like more than the entire legal chronology: the Leningrad computer “Hobbit,” with its Russian-English keyboard, was shipped abroad in small batches—including to the homeland of the original, the United Kingdom. A machine with no rights holder, assembled in a cooperative, went off to be sold in the place where all the rights to it were very much in force.