Once Upon a Pocket: Programmable Calculators from the Late 1970S and Early 1980S and the Social Networks Around Them

[3B2-9] man2012030055.3d 3/8/012 15:5 Page 55

Once Upon a Pocket: Programmable Calculators from the Late 1970s and Early 1980s and the Social Networks Around Them

Dejan Ristanovic´ PC Press, Serbia

Jelica Protic´ University of Belgrade, Serbia

Programmable pocket calculators of the mid-1970s opened up a new segment of the personal computing devices market. Calculator users established clubs, magazines, and conferences, and their interaction with manufacturers shaped the products’ further development. This article explores one of the understudied roots of personal computing, through the evolution of the user communities formed around the TI-59 and HP-41C calculators.

The development of integrated circuits and marketing practices de novo.3 With program- the miniaturization of components in the mable pocket calculators, users were the pre- early 1970s made it possible to construct dominant, driving force behind software small and affordable computers. The long production; they established clubs, commu- road that led to personal computers com- nicated through magazines, and joined social mencedonseveralparalleltracks.Many networks that exceeded geographical limits. attempts to record these events started with Manufacturers welcomed this grassroots the Altair project in 1974, which Paul Ceruzzi movement that helped them enrich their called the annus mirabilis (wonderful year) software base and provide additional new of personal computing.1 Specifically, Intel user support. This article primarily focuses announced the 8080 microprocessor in the on the origins, practices, and accomplish- summer of 1974; in July Radio-Electronics ments of those programmable calculator described the Mark-8; and in late December, user communities. Popular Electronics subscribers received their January 1975 issue with an Altair prototype The Beginning of Programmable on the cover. Before all these events, in Janu- Pocket Calculators ary 1974, Hewlett-Packard introduced the The first pocket calculators were not pro- HP-65 programmable calculator, advertising grammable,4 but the Sharp EL-8 (introduced it as a personal computer,2 which according in 1970) and scientific calculators such as to Ceruzzi is possibly the first use of the the HP-35 and the TI-2500 Datamath were term ‘‘personal computer’’ in print. This useful enough to make the slide rule obsolete, launched a series of events not commonly thus replacing one of the main engineering studied in the history of computing. tools at the time.5 Recognizing this milestone, The evolution of programmable calcula- IEEE in 2009 awarded HP the prestigious IEEE tors is closely related to the development of Milestone in Electrical Engineering and Com- dedicated software. Martin Campbell-Kelly puting Award for its HP-35 scientific calcula- claimed that the personal computer software tor.6 Further advances in chip density and industry had almost no connection with the system software led to more powerful pocket preceding software industry, and therefore it calculators with memory registers, trigono- established most of its development and metric, statistical, and/or business functions.

IEEE Annals of the History of Computing Published by the IEEE Computer Society 1058-6180/12/$31.00 c 2012 IEEE 55 [3B2-9] man2012030055.3d 3/8/012 15:5 Page 56

Once Upon a Pocket: Programmable Calculators from the Late 1970s and Early 1980s

allowed to access a university’s mainframe, the use was restricted and there was no inter- activity.1 A programmable calculator was a truly personal device; users could carry it in their pockets and use it all day (and night). Personal computers, such as the Apple II, Commodore PET, and Tandy TRS-80 (all introduced in 1977), were also a possibility, but the first personal computers cost between US$600 and US$1,300. Introduced the same year, the TI-59 only cost US$300. The difference in price, portability, and numerical precision, together with the fact that personal computers had full-sized qwerty keyboards and high-resolution ab graphic modes suitable for gaming, meant Figure 1. The HP-41CV and TI-59 programmable the two types of machines attracted different calculators. Due to its alphanumeric capabilities, categories of users. A new software industry 3 (a) the HP-41CV had fewer keys but more func- emerged from the early personal computers, tions than (b) the TI-59. The TI-59’s advantages with many companies selling system soft- included numerical precision and simpler access ware, productivity applications, and games. to synthetic functions. Programmable calculators took a different path. According to one TI survey,8 27 per- cent of TI-59 users were engineers, 61 per- Programming, the obvious next step, was a cent purchased their TI calculator for way to replace long, repeating sequences company or business use, and the rest bought with the press of a key. it for personal and educational use. Programming was at first rudimentary be- Researchers in many disciplines discov- cause the sequence of keystrokes was stored ered programmable calculators were valu- in the calculator’s memory. Even the term able, yet relatively inexpensive tools for ‘‘programming’’ was frequently substituted scientific calculations. Results were reported with the term ‘‘learning’’—the calculator in major scientific journals, in papers often ‘‘learned’’ how to solve a problem by per- containing program code for the TI-59 or forming a series of program steps. The only HP-41C. The research areas ranged from aero- instructions exclusively related to program- space9 and electronic systems10 to antennas ming were LRN (start ‘‘learning’’), R/S (run/ and propagation.11 Even medical research stop toggle), and RST (reset the program utilized these programmable calculators; in counter). Program editing was not particu- one example, bacteria identification was per- larly user friendly; the SST (single step) key formed with a calculator instead of the code wasusedtoreadtheprogramstepbystep, book.12 and instructions were displayed as numeric codes. Main Competitors From 1974 until 1979, HP and Texas Because of their relatively long time on the Instruments by turns introduced program- market and enthusiastic user communities, mable calculators, starting with the HP-65 yet not to ignore our personal microhistory, and TI SR-52.7 These models introduced this article is mostly devoted to two compet- flow control instructions, loops, and indirect ing calculators that dominated the market in addressing, and their program memory ca- the late 1970s and the early 1980s: the TI-59 pacity was approximately 200 bytes. Both cal- and the HP-41C (see Figure 1). The TI-59 was culators had an integrated magnetic card introduced on 24 May 1977 as the successor reader so users did not have to enter the pro- to the SR-52, which had a strong user base grams each time they turned on the calcula- and even a dedicated monthly magazine, 52 tor. An optional 20-column thermal printer Notes, published by Richard Vanderburgh. was introduced for the SR-52. The HP-41C was introduced on 16 July The first programmable calculators in- 197913 as a successor to the HP-65/67, duced new social networks of users. Main- which also had a strong user base and a dedi- frames were out of reach for average cated magazine, 65 NOTES, published by engineers, and even when students were Richard Nelson.

56 IEEE Annals of the History of Computing [3B2-9] man2012030055.3d 3/8/012 15:5 Page 57

Table 1. Comparative characteristics of the TI-59 and HP-41C.

Texas Instruments TI-59 Hewlett-Packard HP-41C (1979–1985) Characteristic (1977–1983) and HP-41CV (1980–1990) Initial price US$300 US$295/325 Dimensions 16.3 7.3 3.6 cm 14.27 7.86 3.33 cm Weight 240 grams 205 grams Display type LED LCD Numeric precision 13 digits 10 digits ROM 6 Kbytes, 13-bit words 12 Kbytes/15 Kbytes, 10-bit words Memory registers 120 64/320 User memory 960 bytes 560/2,352 bytes Continuous memory No Yes Alphanumeric No Yes Sound No Beeper Magnetic card 2 30 registers 2 16 registers capacity

Although these calculators were similar in available for purchase. The master library many ways (see Table 1), their programming contained 25 programs, with something for differed. TI used an algebraic operating sys- everybody.15 Engineers got matrix operations, tem (AOS), where the arithmetic expressions mathematicians got numeric integration, were entered the way they are written on the business people got annuities calculations, paper—for example, ‘‘1 þ 2 (3 þ 4) ¼’’. HP and home users got unit conversions, calen- used reverse Polish notation (RPN) based on a dar functions, and a simple game, Hi-Lo. four-register stack—for example, ‘‘1 Enter 2 Other modules included Math/Utilities, Ap- Enter 3 Enter 4 þ x þ’’, or more efficiently, plied Statistics, Real Estate, Aviation, Marine ‘‘3 Enter 4 þ 2x1þ’’. RPN gave HP calcula- Navigation, Surveying, and Leisure Library. tors a dose of exclusivity because amateurs The most talked about was the Olympic mod- couldn’t calculate even the simple expres- ule, which helped verify the results at the sions(‘‘Whereisthe¼ key?’’), whereas the 1980 Winter Olympics in Lake Placid.16 The gurus could perform complex math with a programs in library modules were written in minimal number of keystrokes. the same language as user programs stored in TI-59’s RAM. The user could download Texas Instruments TI-59 any program from the module into the RAM The TI-59’s original price was US$300, which memory. Because the programs were written according to the US inflation calculator, is by excellent TI-59 programmers, there was al- equivalent to US$1,080 today. According to ways something to learn from them. the TI-59 patent,14 the most important inno- The TI-59 instruction set was limited to vation was its solid-state software module 100 instructions, which was enough for com- slot. Modules addressed the common weak- plex commands such as statistical functions, ness of early programmable calculators—the branching, calling subroutines, and indirectly limited amount of RAM, which was insuffi- addressing memory registers. To allow for (di- cient for storing multiple programs. Those rectly) addressing up to 100 memory registers, programs could have been stored in ROM some instructions were designed to occupy memory, but that would have made the cal- two or even three program steps. For example, culator specialized for a specific type of pro- to store a number in memory register 25, the fessional. Mathematicians, civil engineers, instruction STO 25 was used and coded as and electrical engineers, to name just a few 42 25. One prefix, 69, was entitled OP and examples, require different programs for used for some rarely used operations, so it their everyday work. Loading programs was possible to introduce another 100 instruc- from magnetic cards prior to use was slow tions; only 40 of them were actually used. and inconvenient, so the TI-59 introduced a replaceable 5-Kbyte ROM module. TI-59 Programs and Community With each TI-59, the user received a mas- TI’s software strategy relied on library mod- ter library module, and other modules were ules targeting groups of professionals, but

July–September 2012 57 [3B2-9] man2012030055.3d 3/8/012 15:5 Page 58

Once Upon a Pocket: Programmable Calculators from the Late 1970s and Early 1980s

even more specialized programs were neces- Some of the important TI-59 related publi- sary. Therefore, TI established a Software Ex- cations include the following: change Library. The library’s initial content consisted of approximately 200 programs TI Software Exchange, a Belgian software li- written by TI’s programmers.17 With an brary, was an important channel for the annual US$15 membership, subscribers international distribution of TI-59 pro- received three programs of their choice. grams, because PPX didn’t accept subscrib- Each additional program cost US$3. These ers from outside the US, Canada, and amounts covered the nominal costs of copy- Mexico. TI SOFT also published a quarterly ing the documentation and the listings and newsletter edited by Thomas Coppens. mailing all that to the purchaser. Programbiten was a Swedish quarterly mag- TI encouraged users to submit their own azine founded in 1980 and edited by Lars programs, so the growth of the library Hedlund and Bo Nordlin. The publication depended on user input. These programs had a carefully designed layout and were evaluated by TI’s staff and accepted to looked like a magazine, rather than a the library only if they were deemed to be club newsletter. functional and useful. The authors were not L’Ordinateur de Poche, a neatly printed paid for accepted programs, but they were French magazine, covered the TI-59, HP- awarded two other programs from the library. 41C, and Japanese pocket computers. This was somewhat similar to open source Galaksija,aSerbianpopularsciencemaga- software,18 which came later, with one signif- zine, established a programmer’s club and icant difference: users wrote (free) programs in a library with more than 500 programs, order to receive other (free) programs. Over mostly for the TI-59 and HP-41C calcula- the years, the library grew to 3,000 programs tors. The first computer magazine in the covering different areas—mostly engineering, former Yugoslavia, Racˇunari (The Com- 24 math, and utility.19 From January 1977 to De- puters), grew out of that club. cember 1982, TI published a bimonthly newsletter, Professional Program Exchange (PPX). TI-59 Secrets The most important TI-59 related publica- In addition to writing math and engineering tion was TI PPC Notes.20 This newsletter, to- programs, club members mostly talked gether with the TI Programmable Calculator aboutsecretcommandsthatenabledtheir Club, was founded in January 1980 by Maur- machines to do something the manufacturer ice E.T. Swinnen. Although Swinnen was a TI hadn’t foreseen. Although the TI-59 was Professional Productivity Program seminar programmed in a relatively low-level script, teacher,21 the club was independent from no interface to the machine language pro- TI. Beginning in April 1981, a big LRN logo gramming was provided, so users had to was printed on the cover, so many subscrib- rely on operating system bugs to access hid- ers called the newsletter Learn. den features. The club’s purpose, as stated by the founder, was ‘‘to disseminate member-generated soft- HIR Registers. The first trick came from ware, programming aids and tricks, to report analyzing the instruction set. Some instruc- on hardware modifications, new hardware, to tions consisted of two program steps—the tell about books, programs and other material first was the instruction code, and the second related to programmable calculators that has was the address part. For example, the pro- been published, and to report on news about gramRCL82x2 brought the content of our beloved machines.’’22 The club was loosely data register 82 and calculated its square. organized and activities centered around the The program was coded as 43 82 33, where newsletter, which was about 16 pages per 43 was the code for RCL instruction and 82 issue and published 10 times a year. The an- was the address. If the user deleted the first nual subscription rate was US$20. code, 43, 82 became the instruction code, TI PPC Notes became the focal point for taking 33 as its address. According to the other TI-59 related clubs and publications owner’s manual, 82 is not a valid code, but from all over the world. Being a polyglot (flu- when the program was listed on a PC 100C ent in seven languages), Swinnen was able to printer, HIR 33 was printed, HIR being the translate results from other countries and new, secret instruction. publish them in the Notes.23 By the end of In addition to 100 data registers, the TI-59 1982, PPC Notes had a circulation of approx- had eight internal registers used for advanced imately 4,000. arithmetic calculations, statistic functions,

58 IEEE Annals of the History of Computing [3B2-9] man2012030055.3d 3/8/012 15:5 Page 59

and printing. Using a hierarchical internal Graphics Mode. The PC100C printer was a register (HIR) instruction, it was possible to character-oriented device, so it was not possi- access internal registers and use them as addi- ble to plot high-resolution graphics. How- tional storage space. Discovered by Heinrich ever, Michael Sperber and his colleagues Schnepf in September 1977,25 HIR registers discovered that the h25 (h stands for hex) became well-known and were routinely used code halted the printing operation so only to free conventional registers for data storage. the top three pixels of the characters printed.32 By carefully choosing characters, Fast Mode. The TI-59’s fast mode was the programmers were able to design fascinating most controversial discovery, letting users output (for that time, anyway). run programs at double speed. Soon after the TI-59 became available, users realized Hewlett-Packard HP-41C that the code executed from a library mod- To compete with the popularity and sales of ule ran significantly faster than the same the TI-59, Hewlett-Packard needed to offer a code when downloaded into the user mem- significant improvement over its top model ory.26 It turned out that internal arithmetic HP-67 to stay competitive. HP’s answer was operations were executed with numbers in the HP-41C alphanumeric calculator with exponential format. To format the trace on continuous memory,33 the contents of which the printer, programs running in the main were retained even when the calculator was memory converted the virtual display to turned off.34 The initial review published in standard floating-point format and back the BYTE magazine claimed that ‘‘The HP- again between two instructions. Because 41C has an array of features that blurs the dis- the module programs were not traceable, tinction between calculators and personal this conversion was unnecessary, letting computers.’’35 HP claimed that the HP-41C module programs execute faster. was designed to be not just a calculator, but In early 1980, Martin Neef discovered the a ‘‘mini personal information center, or sim- sequence of instructions that could trick the ply said, computing system.’’36 The initial calculator into running the user program as price was US$295, which is equivalent to if it is stored in the ROM.27 Fast Mode’s limi- US$900 today. tations were severe, but because the TI-59 was This system’s key elements were four ex- not a particularly fast computing device, dou- pansion ports in the display end of the calcu- bling its speed was considered important. lator. This idea took the concept of TI-59’s library ROMs a few steps further because Keyboard Branching. The TI-59 was not users could insert ROMs with programs into designed for arcade games because the run- these ports as well as RAM modules, a card ning program could not acquire any informa- reader, a printer, a bar-code reader, and tion from the keyboard, and the only active other peripherals. In later years, HP intro- key, R/S, suspended program execution. Mar- duced the HP-IL (interface loop),37 ashort- tin Neef found a way to use R/S not to halt range interconnection bus and predecessor the running program, but to start a ROM pro- to the Universal Serial Bus (USB). HP-IL gram instead.28 Dejan Ristanovic´ found a way allowed devices such as printers, tape readers, to use this feature to control program execu- RS-232 controllers, and floppy disk drives to tion29 and wrote the 5/10 Start and TI-59 be connected to a programmable calculator Super Test programs using this feature. or a desktop, ISA-bus-based PC. HP also introduced new concepts on the Hex Codes. TI-59 data and program storage calculator itself. Early observers were aston- were interchangeable. The user could turn pro- ished by the fact that the HP-41C keyboard gram steps into data registers or vice versa, was far less cluttered than the HP-67.33 eight program steps per register. However, There was only one shift key, whereas previ- the eight program steps took two bits less ous models had two or three. Where did all than the data register, so it was possible to in- the functions go? Actually, the HP-41C’s al- sert ‘‘undefined’’ instruction codes at each step phanumeric capabilities permitted function with a position that was divisible by eight. By names to be keyed in and shown in the dis- varying extra bits, programmers could exe- play. The alpha key toggled the keyboard be- cute some internal instructions. Hex codes tween alphanumeric and classic calculator were discovered by Patrick Acosta30 and modes, and the letters were printed on the researched by Robert A.H. Prins. Ristanovic´ slanted faces of the keys. Because frequently decoded some of the new instructions.31 typing function names was tedious, it was

July–September 2012 59 [3B2-9] man2012030055.3d 3/8/012 15:5 Page 60

Once Upon a Pocket: Programmable Calculators from the Late 1970s and Early 1980s

possible to assign any function to any key, so in Corvallis, Oregon, by HP. UPL formed a users could redesign the keyboard to suit their software catalog that eventually collected ap- personal needs. HP introduced the concept of proximately 2,800 titles for the HP-67 and overlays, thin plastic matrices the user could (mostly) HP-41C.39 The annual library sub- place on the keyboard and glue preprinted scription was US$20,40 and the programs labels on, making it easier to remember cost US$6 each, which included documenta- which function was assigned to which key. tion, listing, and recorded magnetic card(s). The HP-41C used a custom CMOS/SOS Recognizing Europe as an important market, processor called Nut.38 Thecasealsocon- HP also founded the User Program Library tained five RAM chips, each with 16 7-byte Europe (UPLE) in Geneva, Switzerland. registers and three 40-Kbit ROMs (later mod- MostoftheprogramswerewrittenbyHP- els used one 120-Kbit ROM), so the total 41C users all over the world. The review pro- ROM capacity was approximately 12,000 cess was strict—only high-quality programs 10-bit words.34 Four size N alkaline batteries with extensive documentation were accepted. powered the device, which differed from pre- Authors of each accepted program received a vious models that used rechargeable NiCad point certificate, which they could exchange cells. With this change, HP made the device for four programs of their choice from the li- lighter, and a rechargeable battery pack was brary catalog. Library points could also be col- offered later as an option. lected and exchanged for hardware—three HP-41C’s programming language was sim- points for an HP-41C module, seven points ilar to the dialect used on previous HP’s cal- for a bar code reader, 18 points for a printer, culators and, for that matter, similar to the and so on.41 For 100 points, a user could TI-59’s programming language (they even claim the HP-85 home computer, a US$3,250 shared a lot of mnemonics), but alphanu- value at that time. Furthermore, as recognition meric capabilities brought a new level of of ace programmers, a bronze, silver, and gold user friendliness. Programs had descriptive overlay was issued to every person who had names, could prompt the user for input, one, five, and 10 programs, respectively, and could label program output. accepted to the library (see Figure 2). The HP-41C brought memory manage- TheUPLandUPLEsharedanewsletter ment to the programmable calculators mar- called HP Key Notes (HP-65 Key Notes until ket. Unlike earlier models, each instruction 1976). HP Key Notes was distributed all over could consist of many bytes, up to 16 in the world for free. Published four times per rare cases. Instead of the number of instruc- year, each issue had about 20 pages and was tions, the program length was measured in mostly dedicated to using the HP-41C, bytes. It was possible to have many indepen- hints, and tricks. The newsletter reached the dent programs stored in the user memory, height of its popularity in March 1982 and each program could have the same local when it published exclusive footage from a labels, use the same registers, and so forth. NASA space shuttle mission. (Columbia’s When a program was deleted from the mem- astronauts routinely used HP-41C calculators ory, the HP-41C could do defragmentation for various flight-related, radio contact, and called packing. User key assignments were backup operations.42) also stored in the main memory, requiring When Henry Horn (1928–2007), the long- more advanced packing algorithms. time editor of the HP Key Notes, retired in Au- With all these possibilities, HP-41C’s initial gust 1983, HP discontinued the newsletter memory of 63 registers (about 440 bytes) and the UPLE.43 The Corvallis library contin- quickly became insufficient, so the user had ued operation up to 1988, but without a news- to purchase memory modules, increasing letter and with somehow lower ambitions. the memory up to 320 registers (about The user community was redirected to inde- 2 Kbytes). To keep the expansion ports free pendent organizations, especially the PPC. for other peripherals, HP introduced the HP- The PPC grew around its publication, 41CV (V is the Roman numeral symbolizing which in various periods was called 65 five times more memory) on 15 December NOTES, PPC Journal,andPPC Calculator Jour- 1980. An even more powerful version, the nal. The club’s founder Richard Nelson HP-41CX, was introduced in November 1983. explained that PPC didn’t stand for anything special, but it could mean Personal Program- HP-41C Programs and Community mer Club.44 The PPC Journal was published The official source of the HP-41C programs irregularly, usually eight to 10 times per was the User Program Library (UPL) founded year. The standard issue had 32 pages of

60 IEEE Annals of the History of Computing [3B2-9] man2012030055.3d 3/8/012 15:5 Page 61

condensed, printed text and listings. Articles and programs were written by the club’s members and usually left unedited. Articles were reproduced from the submitted print- out, so the issues consisted of many different fonts and layouts, especially after dot-matrix printers became affordable. PPC was a global organization with members in more than 30 countries. The club encouraged members to form local chapters and pursue local activities, especially regular meetings. Chapters received organizational help from PPC but operated autonomously. In 1982, PPC had 43 chapters, 29 in the US and 14 in nine other countries: Australia, Canada,Denmark,France, Spain, Sweden, South Africa, Norway, and the United King- dom.45 Some of the chapters published HP- 41C related newsletters. The most active chapters were in Australia, which published Technical Notes (later HPHH Melbourne) from 1979 to 1993, and the PPC-UK, which began publishing Datafile in 1982. (At the time of this writing, the current issue of Data- file is vol. 30, no. 2, April 2011).

Synthetic Programming Like the TI-59, the HP-41C was a ‘‘closed’’ system that could only be programmed in user code, without access to the machine lan- Figure 2. Hewlett-Packard’s official Users’ Program Library (UPL) pub- guage and internal data structures. Addi- lished a catalog with more than 2,800 top-quality HP-41C programs. tional access to internal structures was Programs were mostly written by users. The author of an accepted required to fully utilize the calculator’s alpha- program received a point certificate that could be exchanged for other numeric capabilities. Alpha string editing was programs or hardware. virtually nonexistent in the main instruction sets, and when observing results from some of the calculator’s irregular states, it was obvi- users access the HP-41C’s internal structures. ous that the character set contained ligatures The next year Wickes introduced the concept that were impossible to type regularly. of byte jumping as a ‘‘poor man’s black In August 1979, John Kennedy and box’’50—that is, using a synthetic key assign- Richard Nelson published the first version ment to enter synthetic functions without of the HP-41C combined hex table,46 demon- having any special program. It took almost strating the use of a complete character set two years to discover the ultimate synthetic and ‘‘unprintable’’ ligatures. The same issue programming tool, the Byte Grabber,51,52 a of the PPC Journal reported a useful bug: indi- synthetic key assignment that stuck the first rectly addressing nonexisting registers byte of a regular instruction into a dummy 704 through 999 would read/write program text line, making the second byte of the in- code interpreted as data.47 The bug let users struction the first byte of a new instruction, enter ‘‘nonexisting’’ instruction codes in the just like the technique of entering ‘‘illegal’’ program memory and study their behavior. instruction codes on the TI-59. In the next issue, HP recognized users’ inter- Wickes wrote Synthetic Programming on the ests by revealing some internal structures of HP-41C, which soon became the ‘‘bible’’ for the HP-41C,48 so the quest started. HP-41C’s power users.53 Shortly after pub- Quite a few articles were published based lishing that book, Wickes (at that time an as- on trial-and-error research. Pieces started to sistant professor of physics at the University fall into place in an article by W.C. Wickes of Maryland) was recruited by HP and be- entitled ‘‘Through the HP-41C with Gun came R&D director, responsible for devel- and Camera.’’49 His Black Box program let oping advanced scientific calculators and

July–September 2012 61 [3B2-9] man2012030055.3d 3/8/012 15:5 Page 62

Once Upon a Pocket: Programmable Calculators from the Late 1970s and Early 1980s

(later) portable computers. Wickes held that make the HP-41C perform even better? The position until he retired in November 2010. initial ideas were published in the PPC Jour- nal60 and the work started, but the PPC PPC ROM Project ROM II was never published. In August 1979, Nelson proposed that PPC members build an 8-Kbyte custom ROM for Decline and Fall of the PPC Empire the HP-41C, containing utility and other pro- In 1982 Sharp introduced the PC 1500 (and grams.54 The idea was ‘‘to provide a data han- later the PC 1500A), a 19.5 8.6 2.55 cm dling structure that may be utilized as pocket computer weighing 375 grams, with subroutine calls to reduce complex and a 65-key qwerty keyboard, a 26-characters long programs to a few steps, taking the op- LCD display, 16 Kbytes of ROM, and 8 Kbytes portunity for a new level in PPC software of RAM. The use of the Basic programming achievement and application.’’ The PPC language made the computer attractive to a ROM Manual states that ‘‘the PPC ROM proj- broad audience.61 However, its horizontal de- ect represents one of those rare occasions sign and awkward way of storing multiple where a group of people join together to ac- programs in memory didn’t make the first complish a work for the primary reason that pocket computers attractive to engineers it’s a good idea.’’55 and students, so PPC might have lasted for The PPC ROM development was docu- years, had it not fallen apart from the inside. mented in the ROM Progress column of the Producing 5,000 PPC ROMs made it diffi- PPC Journal from August 1979 to July 1981. cult for the club to operate informally, so Month after month, PPC members discussed, on 18 January 1982 PPC filed incorporation proposed, optimized, and corrected pro- papers with the state of California as a non- grams, trying to squeeze as many as possible profit corporation, run by a board of directors. into 8 Kbytes of code. Saving space was the Nelson became president of the corporation. imperative, so the alphanumeric prompting However, a publication supporting the de- was kept to a minimum. Therefore, an impor- vices produced by a single manufacturer (in tant part of the project was the high-quality this case, HP) could not obtain tax-exempt user manual, which became a 500-page status. Thus, beginning in May 1982, the book printed in a small font. PPC Journal also published TI-59 related ma- The ROM project disrupted normal opera- terial, mostly reprinted (with permission) tions of the PPC, but when the final product from TI PPC Notes.62 This decision was not arrived in late 1981, it had something to received well by the membership because show for its work. The tiny ROM module there were not many ‘‘bilinguals’’ familiar contained 122 state-of-the art programs56 both with TI-59 and HP-41C programming. and became a necessity for any HP-41C The social networks that emerged around power user. It prompted HP to introduce its these calculators therefore remained only own Extended Functions/Memory Module loosely connected. that provided machine-code instructions for More serious problems came from within string manipulation, thus reducing the need the board of directors. Running the corpora- for synthetic programming.57 The PPC ROM tion, managing employees, starting another found new use as a tool for directly accessing magazine (the PPC Computer Journal), and newly introduced extended memory, de- handling financing increased the tension signed as additional data storage. With some within the board, so on 11 April 1984 Nelson clever manipulation, the extended memory resigned as president and the journal’s editor. could be used as program memory.58 This was probably just a tactical move, but the Because HP required the purchase of at least board quickly elected Emmett Ingram the new 5,000 units in order to produce the modules, PPC president and David White the new jour- PPC allowed each member to purchase two nal editor. During the next few months, PPC PPC ROM modules for US$95.59 The idea was was in the state of turmoil. As stated in a to let members keep one module and sell or seven-page letter to members in August give away the other, thus making the club 1984,63 the board had to hire a security more popular. After the initial purchase, the guard to keep Nelson out of the clubhouse ROM was sold to new members for US$95 each. during regular Friday night meetings. The successful completion of this project Nelson eventually started a new publica- naturally made the members ask for more. tion, the CHHU Chronicle, which (pro- Why not produce a PPC ROM II, with nounced ‘‘chew’’) stands for the Club machine-language programs that would of Hewlett-Packard Handheld Users. The

62 IEEE Annals of the History of Computing [3B2-9] man2012030055.3d 3/8/012 15:5 Page 63

Chronicle was an amateur publication but Technology (MIT) in the 1950s and 1960s67 could not reach a critical number of members andfollowedin1975bytheHomebrew to support a full-time editor. Only 11 issues Computer Club in the Silicon Valley.1 were published, and it folded in early 1986. To understand the difference between the The PPC Journal lasted a year longer but was evolution of the software industries for early struggling for content; it finally folded in home computers and programmable calcula- May 1987. HP pocket calculator users lost a tors, we should consider Sherry Turkle’s valuable source of information as the need claim that when the early owners of home for information increased after HP revamped computers bought their machines, their soft- the scientific RPN calculators by introducing ware and hardware were in a state of disequi- the HP-28S in 1987 and the HP-48S in 1990. librium.68 Although the new Apple II, PET, or At that point, information exchange occurred TRS-80 could do virtually nothing useful at regular conferences, through books, and without application software, programmable (later) over the Internet. pocket calculators could be used for its main The TI-59 went down much less spectacu- purpose, numerical computation, and their larly, mainly because TI didn’t follow it with manufacturers provided necessary programs a successful model. The successor, TI-88, was in the form of application modules. Even an alphanumeric calculator with an LCD, when programming was a necessity, it did faster processor, up to 3-Kbyte memory, and not require learning a language—elementary better instruction set.64 It could easily com- programs were simply a series of keystrokes. pete with the HP-41C, but not with the When users required more, they sought it next generation of HP’s graphic calculators from one another. equipped to solve symbolic equations, start- Ceruzzi claimed that HP and TI were ing with the HP-28S. After a substantial unprepared for the mass sales of their calcula- R&D expenditure, TI cancelled the project tors.1 They sold the machines as commodities and shifted its priorities to a family of hand- and could ill-afford a sales force that could held computers (such as the CC-40) and its walk a customer through the complex learn- desktop Basic computer, the TI-99/4A.65 ing process needed to get the most out of Without a star calculator, user clubs stag- one. Instead, the manufacturers helped nated. In January 1983, TI discontinued shape user communities. TI initiated Profes- PPX and that same year terminated the pro- sional Productivity Program seminars, in duction of the TI-59. PPC Notes continued which engineers and technicians were trained to publish under a new editor, Palmer O. to use the TI-59 (the price of the user course Hanson, and lasted for 14 years, until No- was US$1,200, three times the calculator’s ini- vember 1991. European clubs mostly stopped tial price).21 HP’s developers and other experts publishing in 1984. were regular keynote speakers at the annual user conferences hosted by the PPC (later The Role of Users HHC, http://hhuc.us), from Santa Clara, Cali- As Campbell-Kelly and William Aspray fornia, in 1979, to Fort Collins, Colorado, in claimed, ‘‘with the arrival of consumer- 2010. More importantly, by establishing user oriented machines in 1977 the market for libraries, both HP and TI shaped user com- applications software took off—applications munities that could provide advanced soft- software enabled a computer to perform use- ware and support for newcomers. ful tasks without the owner having to pro- Because the calculators were sold in many gram the machine directly.’’66 This claim countries, newsletters and magazines were applies to the Apple II, Commodore PET, the most effective form of communication and Tandy TRS-80. Although the HP-65, TI- between users. The groundwork was again 59, HP-41C, and other programmable calcula- laid by the manufacturers, but their editions, tors could be categorized among the first con- the TI PPX Exchange and the HP Key Notes, sumer-oriented personal computers, their were insufficient. Users formed clubs and software development took a different path. published their own newsletters. Between Because no companies sold software for 1977 and 1984, the two most important mag- these calculators, the payment-by-royalties azines, TI PPC Notes and PPC Journal,pub- model was not implemented and few programs lished about 30,000 pages. Most of the were actually sold for profit. The software was articles were accompanied by programs mostly written by users and distributed to ready to be keyed into the reader’s calculator users for free, similarly to the concept origi- to solve math and engineering problems. nated at the Massachusetts Institute of Game programs were occasionally published,

July–September 2012 63 [3B2-9] man2012030055.3d 3/8/012 15:5 Page 64

Once Upon a Pocket: Programmable Calculators from the Late 1970s and Early 1980s

but the games did not shape the pocket cal- the main source for HP-41C emulators writ- culator market as they did with home com- ten for different platforms, from Windows puters3 because the TI-59 and HP-41C and Mac OS X to iOS and Android. The displays were unsuitable for gaming. site’s archive DVD includes scanned HP-41C Unlike the computer magazines, TI PPC documentation, numerous books, and pro- Notes and PPC Journal didn’t publish adver- grams. Usenet newsgroups also still host tisements or color pictures; they consisted HP-41C-related discussions. mostly of small-print text, written by users The major TI-59-related publications, PPX, and often not even touched by the editor. Be- 52 Notes,andTI PPC Notes, are available online cause neither the TI-59 nor HP-41C sup- at www.ti59.com (which is operated by Rista- ported program compilation, the source was novic´) and www.rskey.org (which is operated always open, so users were judged by the by Viktor Toth). Various documentation is quality of their code, and programs were also available on www.datamath.org, operated often further optimized by other users, fol- by Joerg Woerner. A TI-59 emulator running lowing the early hacker culture and the under all versions of Windows was written resulting philosophy, formulated later by by Miroslav Nemecˇek. Another version was Steven Levy as the Hackers Ethic.67 Ceruzzi written by Alan Zanchetta (www.zanchetta. called early programmable calculator users net). An online emulator Vertigo (www. ‘‘the intellectual cousins of the students in n3times.com/vertigo), a PocketPC version the Tech Model Railroad Club.’’1 The impor- TI-59ce, and an Android version ti5x written tant difference is that PPC users were not by Lawrence D’Oliveiro are also available. confined to MIT buildings;26 they spread the Hackers Ethic in many countries. Conclusion Both HP and TI recognized the impact of The TI-59 and HP-41C were powerful pro- user communities on the market success of grammable calculators designed in the late their calculators. With each HP-41C sold, 1970s, aimed primarily at scientists and engi- HP delivered a card that directed the buyer neers. Although different in many aspects, to the PPC for additional support and soft- they shared two important concepts: pro- ware.69 TI produced a separate brochure gramming was done by memorizing series with a list of user clubs all over the world, of keystrokes, and libraries of programs were sending it to each user who asked questions distributed as ROM modules. Attempts were about the calculator’s advanced features and made to port programs from one platform programming. The users’ needs, expressed to another: TI produced an RPN Simulator in some of the articles published in news- ROM, enabling the transcoding of programs letters, shaped the development of these written for earlier HP calculators. Ristanovic´ small systems; applications packs, modules, wrote a TI-59 to HP-41 compiler,71 enabling and peripherals, produced by TI and HP, fol- compilation of TI-59 programs into HP-41C lowed some of the club’s projects. user code and into the HP-41C’s memory. As Christina Lindsay noticed,70 atechnol- However, the compatibility and overall use- ogy’s disappearance from mainstream public fulness of these attempts is debatable. view is not necessarily the end of its life. The Similar to early personal computers, the TI- users stay even after the product is discontin- 59 and HP-41C’s small RAM was a serious lim- ued. The user networks formed in the early itation for programmers. Every program step 1980s still exist, and today the Internet is the was precious, so it was common for a program- main forum for information exchange. The mer to spend hours trying to save a few bytes. HP-41C’s network is stronger, in part because Developers often effectively increased data HP later produced successful programmable cal- storage by packing several values into the culators from the HP-48/49 series. Most of the same register. Given all these obstacles, fasci- materials published in the earlier days—22,300 nating programs were written for these small pages spanning from 1968 to 2010—are avail- computers, including chess programs.72,73 able on a PPC DVD produced by Jake Schwartz As historians have recognized in the his- (see www.pahhc.org/ppccdrom.htm). The tory of computer systems in general, the materials were scanned in at a moderate reso- users ‘‘refined, reshaped or (re)programmed lution but not with optical character recogni- systems to meet various purposes, needs, tion, so searching is not possible. and desires.’’74 The concept of users libraries Programs and materials about the HP-41C controlled by a computer manufacturer ree- are available at www.hp41.org, a website merged in 2008 with the App Store, a digital operated by Warren Furlow. This site is also application distribution platform for iOS

64 IEEE Annals of the History of Computing [3B2-9] man2012030055.3d 3/8/012 15:5 Page 65

maintained by Apple. Other mobile device as an Alternative to Code Books,’’ J. Clinical manufacturers have followed suit, establish- Microbiology, vol. 5, no. 2, 1982, pp. 332–334. ing the Android Market (now Google Play), 13. ‘‘HP Creates A Whole New Standard,’’ HP Key Windows Phone Marketplace, and so on. Notes, vol. 3, no. 3, 1979, p. 1. The principal difference is these platforms 14. S.W. Poland, Microprocessor System Having High follow the payment-by-royalties model, Order Capability, US patent 4153937, Patent whereas authors of TI-59 and HP-41C pro- and Trademark Office, filed 1 Apr. 1977. grams were not compensated when other 15. Master Library: Using the Power of Your Solid users purchased their programs. State Software Module, Texas Instruments, 1977. The need for quality programs and the de- 16. ‘‘TI-59 Goes to Lake Placid,’’ PPX, vol. 4, no. 1, velopment of synthetic programming tech- 1980, p. 1. niques led to the formation of strong user 17. ‘‘SR-52/TI-59 Conversion,’’ PPX, vol. 1, no. 6, communities, spread over many countries. 1977, p. 5. The principal means of information exchange 18. I. Svetel et al., ‘‘Prompt System Redesign: Shifting were newsletters and books in the pre-Internet to Open Source Technology to Satisfy User Require- early 1980s. Users wrote tens of thousands of ments,’’ Computer Science and Information Systems pages and spent countless hours researching (ComSIS), vol. 7, no. 3, 2010, pp. 441–457. and writing programs without any monetary 19. ‘‘The G Update,’’ PPX, vol. 5, no. 5, 1981, p. 1. compensation. These networks still partially 20. M. Swinnen, ‘‘Welcome,’’ PPC Notes, vol. 5, exist, and a large percentage of published ma- no. 1, 1980, p. 1. terial is available in electronic form. 21. ‘‘Professional Training on the TI-59,’’ PPX, vol. 3, no. 6, 1979, p. 2. References and Notes 22. M. Swinnen, ‘‘Editorial,’’ PPC Notes, vol. 6, no. 1, 1. P.E. Ceruzzi, A History of Modern Computing, 1981, p. 1. 2nd ed., MIT Press, 2003. 23. M. Swinnen, ‘‘Editorial,’’ PPC Notes, vol. 7, no. 9, 2. C.C. Tung, ‘‘The ‘Personal Computer’: A Fully 1982, p. 1. Programmable Pocket Calculator,’’ Hewlett- 24. J. Protic´ and D. Ristanovic´, ‘‘Building Computers Packard J., vol. 25, no. 9, 1974, pp. 2–20. in Serbia: The First Half of the Digital Century,’’ 3. M. Campbell-Kelly, A History of the Software Computer Science and Information Systems Industry, MIT Press, 2003. (ComSIS), vol. 8, no. 3, 2011, pp. 549–571. 4. E. Swartzlander, ‘‘Generations of Calculators,’’ 25. H. Schnepf and R. Vanderburgh, ‘‘HIR Operations: IEEE Annals of the History of Computing, vol. 17, A Major Discovery,’’ 52-Notes, vol. 2, no. 9, 1977, no. 3, 1995, pp. 75–77. p. 1. 5. O. van Poelje, ‘‘The Demise of the Slide Rule,’’ 26. T. Ferguson, ‘‘CROM Vs. User Code Execution 2001; www.rekeninstrumenten.nl. Speed,’’ 52-Notes, vol. 2, no. 12, 1977, p. 2. 6. ‘‘HP-35 Scientific Calculator Awarded IEEE Mile- 27. ‘‘Quirks,’’ PPC Notes, vol. 5, no. 6, 1980, p. 4. stone,’’ 14 Apr. 2009; www.hp.com/hpinfo/ 28. M. Neef, ‘‘Branching from the Key Board During newsroom/press/2009/090414a.html. Program Execution,’’ PPC Notes, vol. 5, no. 7, 7. E. Swartzlander, ‘‘Generations of Calculators,’’ 1980, p. 11. IEEE Annals of the History of Computing, vol. 19, 29. D. Ristanovic´, ‘‘Branching from the Keyboard no. 1, 1977, pp. 74–75. During Program Execution,’’ PPC Notes, vol. 6, 8. ‘‘Survey Results,’’ Professional Program Exchange no. 9, 1981, p. 31. (PPX), vol. 5, no. 1, 1981, p. 2. 30. P. Acosta, ‘‘Hexadecimal Keycodes,’’ PPC Notes, 9. V.G. Hansen, ‘‘Simple Expressions for Determin- vol. 6, no. 9–10, 1981, p. 14. ing Radar Detection Thresholds,’’ IEEE Trans. 31. D. Ristanovic´, ‘‘Hex Codes,’’ PPC Notes, vol. 7, Aerospace and Electronic Systems, vol. 18, no. 4, no. 6, 1982, p. 8. 1982, pp. 510–512. 32. M. Sperber, ‘‘Graphics Mode,’’ PPC Notes, vol. 6, 10. R. Pinkham and D.F. Anderson, ‘‘A Technique for nos. 4–5, 1981, p. 3. Modeling p-n Junction Depletion Capacitance of 33. ‘‘HP Announces HP41C,’’ PPC J., vol. 6, no. 4, Multiple Doping Regions in Integrated Circuit,’’ 1979, p. 6. Solid-State Electronics, vol. 29, no. 8, 1986, 34. N.L. Johnson and V.V. Marathe, ‘‘Bulk CMOS pp. 773–777. Technology for the HP-41C,’’ Hewlett-Packard J., 11. M. Burns et al., ‘‘A Comparative Study of Linear vol. 31, no. 3, 1980, pp. 20–22. Array Synthesis Technique Using a Personal 35. B.D. Carbrey, ‘‘A Pocket Computer? Sizing Up Computer,’’ IEEE Trans. Antennas and Propaga- the HP-41C,’’ BYTE, Dec. 1980. tion, vol. 32, no. 8, 1984, pp. 884–887. 36. B.E. Musch et al., ‘‘Powerful Personal Calculator 12. J. Schindler and Z. Schindler, ‘‘Numerical Identi- System Sets New Standards,’’ Hewlett-Packard fication of Bacteria with a Hand-held Calculator J., vol. 31, no. 3, 1980, pp. 3–12.

July–September 2012 65 [3B2-9] man2012030055.3d 3/8/012 15:5 Page 66

Once Upon a Pocket: Programmable Calculators from the Late 1970s and Early 1980s

37. R.D. Quick and S.L. Harper, ‘‘HP-IL: A Low-Cost 65. ‘‘TI Portable CPU Line May Dump Top Calcula- Digital Interface for Portable Applications,’’ Hewlett- tor,’’ Electronic News, 27 Sept. 1982, p. 57. Packard J., vol. 34, no. 1, 1983, pp. 3–10. 66. M. Campbell-Kelly and W. Aspray, Computer: A 38. L. Wilkins and P. Lind, ‘‘M-Code, Hardware and History of the Information Machine, Westview Software,’’ Proc. PPC Southwest Conf., PPC, Press, 2004. 1983, pp. 22–43. 67. S. Levy, Hackers: Heroes of the Computer Revolu- 39. Users’ Library Series 40 Software Catalog, Hewlett- tion, anniversary ed., O’Reilly Media, 2010. Packard, Aug. 1984. 68. S. Turkle, The Second Self, 20th anniversary ed., 40. ‘‘Library Corner,’’ HP Key Notes, vol. 4, no. 3, MIT Press, 2005. 1980, p. 3. 69. H. Horn, ‘‘Where Do I Go From Here,’’ HP Key 41. ‘‘HP-67/97/41 Users’ Library Redemption List,’’ Notes, vol. 8, no. 2, 1983, p. 2. Hewlett-Packard, 1983. 70. C. Lindsay, ‘‘From the Shadows: Users as 42. ‘‘HP-41’s Again Aboard Columbia,’’ HP Key Designers, Producers, Marketers, Distributors, Notes, vol. 6, no. 2, 1982, p. 1. and Technical Support,’’ How Users Matter, MIT 43. ‘‘HP Users’ Program Library Europe to Close,’’ Press, 2005, pp. 29–50. Datafile, vol. 2, no. 3, 1983, p. 31. 71. D. Ristanovic´, ‘‘TI59 to HP-41 Compiler,’’ PPC 44. ‘‘Member Letter,’’ 65 NOTES, vol. 4, no. 10, 1977. Calculator J., vol. 9, no. 6, 1982, pp. 27–29. 45. ‘‘PPC Chapter List by Country,’’ PPC Calculator 72. M. Sperber, ‘‘CHESS 2.1,’’ PPC Notes, vol. 7, J., vol. 9, no. 6, 1982. no. 7, 1982, p. 20. 46. ‘‘HP-41C Combined Hex Table,’’ PPC J., vol. 6, 73. C. Roeltgen, ‘‘Chess for HP-41,’’ PPC Calculator no. 5, 1979, pp. 22–23. J., vol. 9, no. 4, 1982, p. 78. 47. B. Danby, ‘‘B2,’’ PPC J., vol. 6, no. 5, 1979, p. 28. 74. J.R. Yost, ‘‘From the Editor’s Desk,’’ IEEE Annals 48. ‘‘HP-41C Data and Program Structure,’’ PPC J., of the History of Computing, vol. 33, no. 2, vol. 6, no. 6, 1979, p. 19. 2011, pp. 2–3. 49. W.C. Wickes, ‘‘Through the HP-41C With Gun And Camera,’’ PPC J., vol. 6, no. 8, 1979, p. 27. Dejan Ristanovic´ is the 50. W.C. Wickes, ‘‘Byte-Jumping or The Poor Man’s founder and editor in chief of Black Box,’’ PPC Calculator J., vol. 7, no. 4, the computer magazine PC 1980, p. 26. Press and the cofounder of 51. E. Gostelli, ‘‘Generalized Fn Byte Maskers,’’ PPC SezamPro, one of the leading Calculator J., vol. 8, no. 5, 1981, p. 11. Internet providers in Serbia. 52. K. Kendall, ‘‘Bare Bones Prefix Masker,’’ PPC He is the author of the first Calculator J., vol. 8, no. 6, 1981, p. 40. publication about personal 53. W.C. Wickes, Synthetic Programming on the HP- computers in the former Yugoslavia, published in 41C, Larken Publications, 1980. December 1983, and cofounder of Sezam, the big- 54. R. Nelson, ‘‘PPC Custom ROM,’’ PPC J., vol. 6, gest bulletin board system (BBS) in the former no. 5, 1979, p. 27. Yugoslavia. In the early 1980s, he was an active 55. PPC ROM User’s Manual, PPC, 1981. member of many programmable calculator user 56. ‘‘ROM Routines Listed by Documentation Coordi- clubs and wrote numerous TI-59 and HP-41C pro- nator,’’ PPC ROM User’s Manual, PPC, 1981, p. VII. grams published in the clubs notes. Contact him 57. ‘‘HP Extended Function/Memory Module,’’ PPC at [email protected]. Calculator J., vol. 8, no. 6, 1981, p. 36. 58. R. Atwood, ‘‘A Program File Can Run,’’ PPC Cal- culator J., vol. 9, no. 3, 1982, p. 24. Jelica Protic´ is an associate 59. ‘‘Member Letter,’’ PPC Calculator J., vol. 7, professor of computer engi- no. 10, 1980. neering and informatics in 60. L.A. Wilkins, ‘‘HP-41 Machine Language Devel- the School of Electrical Engi- opment Lab - MLDL,’’ PPC Calculator J., vol. 9, neering at the University no. 3, 1982, p. 37. of Belgrade. Her research 61. D. Ristanovic´,, ‘‘Zagonetke Sharp-a PC1500’’ [Sharp interests include distributed PC 1500 Puzzles], Racˇunari u vasˇoj kuc´i [Computers systems and all aspects of in Your Home], vol. 1, Dec. 1983, p. 80. computer-based quantitative performance analysis 62. ‘‘Member Letter,’’ PPC Calculator J., vol. 9, no. 4, and modeling. With Milo Tomasˇevic´ and Veljko 1982. Milutinovic´,shecoauthoredDistributed Shared 63. ‘‘Member Letter,’’ PPC Calculator J., vol. 11, no. 7, Memory: Concepts and Systems (IEEE CS Press, 1984. 1997). Protic´ has a PhD in electrical engineering 64. M. Swinnen, ‘‘TI-88—Read all About it,’’ PPC from the University of Belgrade. Contact her at Notes, vol. 7, nos. 7–8, 1982, p. 1. [email protected].

66 IEEE Annals of the History of Computing