CONDUCTED BY GREGG WI LLIAMS AND ROB MOORE

STORY

THEPA RT 2: MORAPEPL HISETO RY AND THE APPLE III

ast montli. Steve talked about liisback­ pany. Lately he's written the Macin­ ground. tlie evolution of tlie tosh word processor-MacWrite. He's and II. and tlieearly days of tliecom­ done a lot for the company, and he's pany. In tliis part. tlie conversation switclies used Sweet-16 in several things he's toL various aspects of tlieApple II design, later done. personal liistory. and Steve's tliouglits about tlie personal computing industry . THE DISK DRIVE : Can you tell us a little about how you SWEET·I6 came up witli the Apple II disk drive and now you ended up picking your form of group­ BYTE: One of tliemore interesting tliingsin coded recording? tlieApple II ROM was your 16-bit pseudo­ WOZNIAK: The disk design was my macliinecalled "Sweet-16:· How did you come most incredible experience at Apple up witli tliat? and the finest job I did. I never really WOZNIAK: While I was writing my knew what a disk controller was or BASIC. I had been thinking about what it had to do. But at Hewlett­ ways to save code. There were several Packard I had looked through a places where I had to handle 16-bit Shugart manual to see what signals pointers with an 8-bit processor. and were used and what they did. There An interview that was pretty awkward. were signals to make the head step in So I decided to write a little and out and signals to cause magnetic emulator and implement a 16-bit with flux changes. It was similar to audio machine that could interpret pseudo­ recording. and I knew about that. It codes and implement registers 0 to was like a signal on a tape where you 15 in the 6502 base page. It ran about write it and then you read it back. So 30 times as slow as 6502 assembly I figured out a simple little circuit to language. but it saved tons of code write signals at changing rates and every time I used it in a program. read them back. I didn't know how BYTE: Did you actually use it in your In­ disk controllers worked. so I assumed teger BASIC? that I was doing something totally dif­ WOZNIAK: No. I never had the time to ferent. Maybe it wasn't as efficient. reimplement the BASIC to use it. But but at least I could write some data I did use it in later years to write and read it back. things like BASIC renumbering rou­ We ll. Mike Markkula was annoyed tines totally in Sweet-16. It was easy because the was too to mix Sweet-16 code with assembly · slow. He had a favorite checkbook language. program. and it took two minutes to read in the prograin and another two BYTE: lsn' t Sweet-16 still used in Apple (continued) DOS and ProDOS editor/assemblers? WOZNIAK: Ye s. it's used in EDASM !the Gregg Williams is a senior teclinicaleditor at Apple Tool Kit 6502 editor/assembler!. BYTE. Rob Moore is a liardware designer and mostly in the editor portion. Randy frequent contributor to BYTE. Tliey can be Wigginton wrote EDASM. He's worked contacted at POB 372. Hancock. NH here since before we even had a com- 03449.

PHOIDGRAPH COURTESY OF APPLE COMPUTER INC. JANUARY 1985 • BYTE 167 WOZNIAK INTERVIEW

I wrote on the wrong one!" We format. double-density recordings. managed to recover it and actually 'Hobbyists are a tiny BYTE: It sounds like a fascinating part. Do demonstrated it at CES. !:fOU think we' see Apple II owners benefit part of our market, BYTE: Had !:fOU been exposed to group-code from it in an!:fII wa!:f? recording before. or did !:fOU invent !:fOUrs WozNIAK: Well. it's our standard disk independent/!:!? controller now. and it's cheaper than hut they're fa ithful WO ZNIAK: The first .version of the the older design. It's used in the floppy-disk routines did not have Apple to the company.' group coding. I had followed the lie. · BYTE: Could !:fOU useit to get higher-densit!:l Shugart manual. which showed that recording on an Apple II 5 \t.i-inch disk? you had alternate clock bits and data minutes to read in his check files. He WozNIAK: No. because the disk drives bits. so every other bit was wasted. I was complaining about this at a staff themselves aren't certified for double­ couldn't understand why it was ­ meeting, and I mentioned that I had density recording. Yo u need heads essary. but I started that way. this clever little five-chip circuit that with the proper gap, and they're more Then I came up with this idea to use could read and write a . At expensive. coded recording. I knew the technical the time. all the existing floppy-disk rule was that you could only have one BYTE: You're a former hobb!:!ist. Could a controllers were 40 or 50 chips. so I or two zeros in a row. Yo u could have hobb!:!ist bu!:l one of these chips cmd a spec knew there must be something impor­ either 4 or 8 microseconds between sheet and start pla!:fing with it? tant that I wasn't doing. flux transitions. I didn't really know WOZNIAK: I don't think Apple would I went off and tried to figure out what group-coded recording was; I give out the spec sheet. I totally dis­ what it was that I wasn't doing. One just knew that I could fit 13 sectors on agree with that policy because I'm of our technicians had a North Star a disk instead of 10. had to write a very respectful to the hobbyists. system. so I looked through their program to take bytesI off the disk. They're a tiny part of our market. but manuals. I read their schematics and convert them to 5-bit chunks. and re­ they're loyal supporters and faithful figured out what every chip did. And assemble them into 8-bit data . people to the company. If they had a I looked through their listings until I It was a difficult routine to write. It spec sheet. they could start playing understood exactly what they were was about a 20-hour job. and I'd work with it and figure out a lot more in­ doing. through the day for 0 or 12 hours credible things that we never planned I was doing a lot more. I didn't even and I wouldn't quite I get there. The it to be used for-even using it as a have to look at the sector holes. so next day I'd come back and find out communications channel from Apple I could use any disk drive. any floppy that I was starting exactly where I had to Apple. to Macintosh. disk in the world. It was then that I the day before. This went on for There are a lot of great tricks you new I had a really clever design. almost a month. I was not quite get­ could do with that little part. It's a The next week was Christmas vaca­ ting the routines. and we were getting beautiful random 1/0 device that has tion. and I spent the within a month of shipping the disk too many things that have not been entire week. including the holidays. drives. Finally I stayed up all night until taken anywhere. trying to get this disk reading and I got all five routines that had to work writing with a very simple operating together done. So we were able to system. We did the bottom levels of PERSONAL DETAILS ship it the first time with the group­ an in that week. Yo u BYTE: In 1981 !:fOU were in a plane crash coded recording in place. Later. we could type R (for "read") followed by and !:fOU left Apple for a while short/!:! after changed the encoding method and a program name like STARTREK. and that. How long did it take !:fOU to recover from stepped up to 16 sectors. That was it would load STARTREK into memory. the accident? DOS 3.3. We were highly motivated because. WozNIAK: That was in February 1981. at the end of the week. a show called BYTE: The Macintosh uses a custom chip For about five weeks I had a type of the CES !Consumer Electronics Showj called the IWM-Integrated Woz Machine­ amnesia that prevents you from form­ was starting in Las Vegas. and we that does the same sort of recording. Can !:fOU ing any new long-term memories. wanted to go. tell us an!:fthing about that? After I recovered. people would show We worked all night the day before WozNIAK: My design was basically a me pictures of myself in the hospital. we had to show it jthe disk drivej at little sequencer. or state machine. It playing games with my computer with CES. At about six in the morning it used a PROM and a latch and cycled my face all battered up. They would was ready to demonstrate. Randy through various states depending on tell me stories of how I tried to sneak thought we ought to back it up. so we the input data coming off the floppy out of the hospital to visit my wife. copied the disk. track by track. When disk. The IWM takes that design and Candy, or how I went to parties and we were all done. he looked down at adds other features like the ability to rode my motorcycle. didn't remem- them in his hands and said, "Oh. no! go twice as fast-it can also do IBM I (continued)

168 BYTE • JANUARY 1985 WOZNIAK INTERVIEW

ber any of that. I had all of my old . WozNIAK: I was going under an as­ My economics course was interest­ memories. but I'd forget new things sumed name-Rocky Clark-so they ing also. We had a socialist TA !teach­ from one day to the next. Finally I didn't know who l was. I took com­ ing assistant] who taught us that com­ came out of it one night but l never puter science courses. economics. panies made money by cheating the got those memories back. statistics. and a few other courses. consumer. All the kids in the class My computer science courses were BYTE: Why did you leave Apple? thought that companies would make interesting. but I have to criticize them a lot of profit if they could figure out WozNIAK: We had a hundred engi­ a little because they taught only a way to cut the costs of a product neers at that point and l was no specific problems with specific solu­ down. to make it cheap and screw the longer really important to the com­ tions. Yo u spent your time memoriz­ consumer. pany. l didn't want to be a manager; ing standard problems and solutions l contrast that with the way we did was just an engineer. and I wasn't and then tried to recognize variations things at Apple. Every product design Ire ally needed there. But didn't feel of them in the tests. Yo u weren't sup­ decision was based on what con­ comfortable going to Stevel Jobs or posed to explore new avenues or try sumers wanted. what would compete Mike Markkula and saying l wanted to things that nobody else was doing. the best what they would buy. We take off. The plane crash was a good Yo u were only supposed to learn the tried to do what customers wanted. excuse. After five weeks of amnesia. proper answer. They thought that you in our best judgment and give them I simply didn't go back. I decided that could be trained to know all the prob­ high-quality products. if I was going to take a year off I might lems and the standard solutions. So I would stand up in class and as well finish college. It was the hard­ Once you learned them alL you could argue about what the TA was saying. est year of my life. solve them. It was wrong because After a while he started telling me to BYTE: We' ve heard that you went to UC­ they weren't really teaching you to shut up. or that he would kick me out Berkeley and had some run-ins with your in­ solve problems-they taught you to if I interrupted him again. Apple was structors. Could you tell us about that? identify them. the greatest business success in history. but I couldn't tell him who I was.

BYTE: So you came back to Apple after If you think all about a year. What would you say is the most important thing that you've worked on since port expanders are equal ... you came back? WOZNIAK: There isn't too much. When I came back I started getting a little bit involved with this division's man­ HERE'S THNo more scrapsE to seeSW who getsIT to useCH the printer ! . agement but it was unofficial. Official­ first. No more tangles with cables everytimeyou want to connect to peripherals. No more jostling ly I took the title of Engineer. Mostly. heavy equipment around the office. I've stayed involved with the Apple II WOULDN'T YOU RATHER SWITCH because that's where I've got the most THAN FIGHT? to offer. Portal allows you to quickly and easily Because I'm a founder at Apple. I introduce a comprehensive switching could take almost any role I want but system that will handle terminals, printers, modems, CPUs and other I've tried to avoid the newest most RS232 compatible devices. Up to 32 far-reaching projects because there ports can handle 16 simultaneous are other capable people to do them. connections. I try to stick to small projects where SAVING TIME AND MONEY­ can sit down and handle them my­ NOW THAT'S A SWITCH I Portal gives you network capabilities at a self. fraction of the cost. With features such as hunt group (acts much like a multi-line telephone that CURRENT WORK will ensure the user gets through to an open mode) BYTE: Do you have anything interesting and wait queues that will automatically notify user going on now? once the requested peripheral is available. There's more good news and features. Call or write today WOZNIAK: There are not many in­ for complete details. dividual projects in this whole busi­ 150 Metcalfe St. ness. The floppy disk may have been Suite 1 028 , Ottawa the last one for Apple. I have got a few Ontario K2P 1 P1 613-230-1759 projects that I'm working on now. but Po- RI- -:AL they're not all going at once.

170 BYTE • jANUARY 1985 Inquiry 126 Inquiry 107

WOZNIAK INTERVIEW

"How can our department get our six computers and three TOUGHprinters to LOCAL work together efficiently? We NETWORKPROBLEM: also want to be able access outside data services and our to future company LAN." The language Fifth is one of them, but I haven't written it yet. I'd like to combine the best features of BASIC NetCommander is a smart, small Local FORTH. and PascaL and leave out the SIMPLE NETWORKArea Network SOLUTIONmanager. It lets= NetCommander you link worst ones. Yo u could have the formal from four to 40 computers and peripherals - structures of PascaL the immediate­ in any mix of models and makes. A 50K buffer ness of BASIC and the extensibility (expandable to 250K) makes sure that productivity of FORTH. In FORTH or Logo, new is high - keeping fewer printers humming - while words become part of the language. computer and PC users do their thing, without waiting and you can use them immediately. for a printer, modem, or shared disk. Those devices can be It's really helpful for debugging. In specified with names defined by users - and allocated on the Macintosh PascaL you can define a basis of availability and capability. And NetCommander handles procedure and run it immediately. I multiple protocols and different baud rates simultaneously - also want some level of globalness, without modifications to hardware or software. It will also tie into your like BASIC. I don't want to always have company's LAN. The latest in a family of products in use since 1979, to declare a variable before I can use NetCommander is a smart, small, efficient network manager. it. I like variables with scope, but I'd For more information, call or write: like undeclared variables to be total­ ly global. Another project is an operating sys­ Digital ProductsNetComm Inc. The Simple Networkander Solution Company tem, like the one on the Macintosh. 600 Pleasant• Street Wa tertown, 02172 only a bit different and a lot more (617) 924-1680 • Outside• Mass. , ca ll 1-MA800-243 • -2333 relational. And check out our 30-day trial eva luation. I also have a hardware project that I'd like to do. a based around consumer video sources like TVs, videotapes, and videodiscs. It would switch them around. synchronize them. and mix them, sort of like a little home editing studio. I think it's possible because memories are getting so cheap. You could hold a frame from each of your Free video sources in memory and let the software accommodate the sync Yo ur 80-page guCatide to computeralog�supp lies and variations. There are a lot of new chips accessories-including complete available that do NTSC modulation newPacked prod with over uctproducts descri for ,ptions. minicomput ers, and demodulation. so there ought to • and word processors1600- many available nowhere else. be a minimal chip solution. Big special section devoted to new supplies and accessories. My main interest is still the Apple • Comprehensive product descriptions - including more than 475 11-the home computer we started • full-color photos - clearly explain features and benefits. with. I'd like to see Apple do more Easy-to-use cross reference guides to magnetic media, ribbons, with speech. There are some really in­ • and more-along with the industry's most complete cable guide. expensive speech chips now. and Helpful suggestions and tips, ranging from flexible disk that's the way the rest of the personal • care to proper ribbon selection to useful application ideas. computer world is heading. I think we've been deficient in that area. PhIn California,one toll-free call 1-800-547-544l1- 800-547-5444 SPECULAT IONS ------BYTE: Are you thinking about using the new ---•• , 6 5 816 processor for anything? • 1-800-547-5444 r Phone toll fr ee * or send coupon today WozNIAK: We're thinking about it and 1nmao NAME lnmac Calalog Dept. doing some R&D with it but I don't 2465 Augustine Drive COMPANY I know if we'll use it. Anything we do I Santa Clara, CA 95054 has to be compatible with the Apple Please rush my free copy of �AD=O�RES�S______CITY If we found out that the 65816 the lnmac Catalog. I under­ I STATE ZIP PHONE \1. (continued) I stand there is no obligation whatsoever. •1n California, call 1-800-547-544l i

InquiryI t 70 JANUARY 1985 BYTE 171 I • WOZNIAK INTERVIEW

THE

n late 1982. A.P.P.L.E.. a national Apple Scott was camped out in the Apple Ill that's good" or something like that. It users group. offered an astounding value building. forcing the things that had to was not official or formal. Randy went toI its members. For $22.50. A.P.P. L.E. get done to get it finished and ahead and hit on it and gave it to Caii­ members could purchase a fully functional Visi­ delivered on schedule. A.PPL.E. It was sold for $22.50-really Calc-like spreadsheet program called "THE He was in a bad mood. Everywhere cheap. We didn't want any money out Spreadsheet:· At the time. various versions of anyone ran into him in the company. of it. It was just the old Homebrew VisiCalc and other popular spreadsheet pro­ he had a sour face-no laughs. no Computer club spirit of "Give to help grams were selling for between $200 and jokes. And because Randy finished his others." $300. THE Spreadsheet offe red most of the first part of the spreadsheet and I It was a good product. and it had fea tures found in VisiCalc and added a few of hadn't finished my first part. he'd been delayed well over· a year. If Apple its own. What's more. it was coauthored by the always come to me. The way he was didn't want it. they should have just now legendary Steve Wom iak. Mysteriously. talking to me. I was afraid I was going gotten rid of it. and it would have been THE Spreadsheet was available fo r only one to get fired. He was in a bad mood. out a year sooner. As soon as it month. after which it was permanently discon­ It was the scariest time of my life at popped up from Caii-A.PPL.£. and was tinued. Since that time. original copies of THE Apple. I was really getting badly ad­ advertised. Apple got in a real huff Spreadsheet have become one of the few dressed just for not being on time with about it. They forced Caii-A.PPL.£. to legitimate collector's items. In something. I had three more days to only sell it to members. and only one the fo llowing segment. Steve describes how THE go. and I didn't dare run into him once copy per member. So Apple basically Spreadsheet came about. more because I was already way over­ put it out of business-only allowed it due. So I said. ''I've got to get him in to be sold for one more month. WoZNIAK: It started because Mike Scott a good mood for three days:· It's really funny. because how did was having difficulty negotiating with We ll. he's a Star Wa rs fan. He gets the Apple get started? I designed the Personal Software jnow VisiCorpj T- shirts the cast gets before they get Apple while I was at Hewlett-Packard. about VisiCalC. They just wouldn't do them because he knows who prints Hewlett-Packard has a policy of decid­ what he felt was right as far as giving them. So I had a friend of mine call his ing about these things very quickly. Apple breaks on price or doing new secretary saying. "This is George Lucas. They would check with their legal enhancements to it. So he came by one Is Mike Scott in?" Of course he department and other divisions and day and he said to me. "Wo'uldyou like wouldn't leave a phone number. but he decide very quickly. So they gave me to do a VisiCalc?" It wasn't even known said. 'Til call back:· It actually worked. a formal release on it quickly because as a spreadsheet then. Mike was in a good mood for a few they didn't want it. But here was Apple. I was really scared. VisiCalc was the days. and I got my routines done. which didn't want it but wasn't going only spreadsheet for a long while. and Anyway. we finished the spreadsheet. to release it. none of us had really been in business and Apple started looking it over. If you use much Apple II software. before. Wa s it legitimate to go out and Originally it had been defined to be as you'll be surprised when you take the do your own spreadsheet? Or was it accurate as VisiCalc. and we were so spreadsheet that Randy and I wrote equivalent to just copying and ripping much faster than VisiCalc it was and boot it in. It boots in so fast. you somebody off? I didn't want to get ridiculous. The exponential routines can't believe what's going on. close to this whole thing because I that I really wanted to write for so long. BYTE: Is that because there's no copy­ didn't want it to look like I was copy­ because I had a good algorithm. were protection on it? ing somebody else. 30 times faster than VisiCalc's. WozNIAK: No. Believe it or not. I used Mike got Randy Wigginton and All of a sudden we started hiring ex­ those routines that I did for Apple II Randy said. "Sure. I'll do it." I said I'd perts on precision and things like that. Pascal. do the arithmetic routines because and they started redefining the pmject they were general-something that from what Mike Scott had originally BYTE: Yo u. Randy Wigginton. and Guil could be used by any program-and I defined. They added a lot of features Banks were listed as the authors. Who is Guil had some good algorithms that I had that Randy had to implement. This Banks? picked up at Hewlett-Packard that I made life very difficult here at Apple. WoZNIAK: Guil is right here in this build­ wanted to implement. so we finally decided. "No. it won't be ing. I did the boot stuff. and he had So we started working on it and. a a Special Delivery Software product ja written a lot of the other fast disk rou­ little ways in. Randy finished it. He line of software offered by Apple a few tines in there. He did all the routines wrote 4K of code in a couple of weeks. years agoj." that managed the disk and read and and I was shocked that he came in so After a long. long time. Randy went wrote DOS files appropriately without quick: I hadn't even started mine. So. to Caii-A.PPL.E. jthe magazine put out using DOS. I was getting to work on my arithmetic by A.P.P.L.E.J. One night we were out This was basically an all-volunteer routines and my first demo was almost to dinner and he said to Steve. "Well. project that we did all on our own ready but still a few days off. I'm going to sell that to Caii-A.PPL.E." And it was really neat. so we decidedtime. At the time. everyone was trying to or 'Tm going to give this spreadsheet that we'd give it to the rest of the . get the Apple Ill into production. Mike to Caii-A.PPL.£. . Steve said. "Well. world.

172 BYTE • · JANUARY 1985 WOZNIAK INTERVIEW DeSmet wasn't it would be a serious question. properly on an Apple 11 or II+. A lot It's too new a part right now. of good software for the Apple II+ c won't run on the Apple lie. It's an in­ BYTE: How is its performance compared to compatible world. the 68000? 8086/8088Developmen t WoZNIAK: It should be available soon BYTE: How do you think the Apple II fa mi­ ly will be extendedand improved in the future? Package in an 8-MHz version that will beat the FUll DEVElOPMENT PACKAGE pants off a 68000 in most applica­ WozNIAK: There are obvious areas. Full K&R C Compiler tions. and in graphics applications it We're always trying to come up with Assembler, Linker Librarian comes pretty close. Some of the better combinations of features and Full-Screen Editor & Macintosh people might disagree still reduce the cost. We're looking Execution Profiler with me. but there are ways around into improved processors like the Complete STDIO Library (>1 20 Func) most of the problems they see. An 65816 we discussed before. Video res­ Automatic DOS 1.X/ 2.X SUPPORT 8-MHz 65816 is about equivalent to olution is always improving. We're try­ BOTH AND S/W FlOATING POINT a 16-MHz 68000 in speed. and a ing to increase speed and the amount 8087 16-MHz 68000 doesn't exist. of memory in the machine because OVERlAYS it's critical to certain applications. BYTE: The Apple II fa mily has been a great OUTSTANDING PERFORMANCE The IBM PC is very successful. and success. and many innovations nave come First and Second in AUG '83 BYTE it had no competition from Apple for along to extend its life. What things do you benchmarks the last three years because we made think were responsible for its success? sure that it didn't. We would not allow WozNIAK: For three years we were the Apple II to compete in that market one of the biggest business successes for three years. SYMBOLIC DEBUGGER in history. We had three years of fan­ PCSD is the most difficult division Examine change variables by tastic business success. and lately & in the company. !Steve is referring to name using C expressions we've had three years of sort of the Personal Computer Systems Divi­ Flip between debug and display dismal business. We've grown. but any sion. responsible for the Apple II and screen additional revenues have just re­ Apple III products.! We're hamstrung Display C source during execution placed the stock dilution. and the by the need to be compatible: Macin­ Set multiple breakpoints by function price remains about the same. tosh isn't. or line number During that three years there were So we will continue to make im­ two main factors that led to our provements and produce new ma­ success-our floppy disk and VisiCalc. chines. but they'll always be compat­ Out of the original home computers. ible. With the Apple lie we went to DOS LINK SUPPORT which included the TRS-80 and the Uses DOS .OBJ Format 128K. 80-column display. and double­ Commodore PET. ours was the only LINKs with DOS ASM resolution graphics. We came out with one that had enough memory to run Uses Lattice® naming conventions the lie portable. VisiCalc. VisiCalc and the floppy disk· sent this company into the number­ Check: Dev. Pkg 09) one position. APPLE VERSUS IBM 0 Debugger (1(50) 0 We were also very faithful to our BYTE: What did you mean when you said DOS Link Supt (35) D users-we tried to support everybody. · that the Apple II was not allowed to compete

When we changed over to floppy in the IBM market? SHIP TO: ______disks. still supported cassettes WozNIAK: Apple has never really sup­ heavilywe. When we moved up to the ported the Apple II in the business Apple II+ with floating-point BASIC market. If you walk around a trade built in. we still supported the original show and look at the software running ------ZIP__ _ Apple II. on the IBM PC. you'll see that most . Lately our strategies seem to be of it is a step above what's possible i changing. When we come up with a on today's Apple II. They have more new enhancement we start moving RAM that's easily addressed and bet­ I away from the prior version much ter access to hard-disk drives. Pro­ I more quickly. This could be harmful grams like 1-2-3 cannot be easily im­ P.O. BOX C to our good relationships with a lot plemented on a 128K Apple II. but CA I Sunnyvale, 94087 of our faithful users. IBM has a capable machine for that I (408) 720-9696 The Apple lie is a good example. level of software. Our machine has to All orders shipped UPS surface on IBM format disks. I Shipping included in price. California residents add Most new software that really uses the be able to address more memory and sales tax. Canada shipping add $5. elsewhere add I features of the Apple lie won't run (continued) $15. Checks must be on US Bank and in US Dollars. I Call 9 a.m - 1 pm to CHARGE by VISNMC/AMEX. l ------JANUARY 1985 • BYTE 173 I WOZNIAK INTERVIEW

won't really be used right away. the Apple II market. After VisiCalc. it 'I have · positive and was perceived that 90 percent of all BYTE: Have you said all that you meant to Apple lis sold were going to small say about where Apple is vis-a-vis IBM and businesses. Only 10 percent were go­ negative feelings about where you personally feel that Apple has ing into this home hobby market that made its mistakes? we originally thought was going to things Apple has done: WOZNIAK: I have both positive and grow to be billions. Originally we were negative feelings about things Apple a home hobby computer. Now. sud­ has done. but I'm always honest. denly. small businesses were buying handle larger disk drives before we There is one real mistake that Apple Apple lis. and they wanted more fea­ can really start to compete with made. in my opinion. and this is very tures-an 80-column display, lower­ equivalent software because IBM subjective. It is symbolic of what could case characters. maybe more graphics beats us in capability today. We need happen to IBM with its PCjr. modes and colors. and more memory. better screen resolution. more mem­ We had become a huge business These were all the things that one ory. and better speed. A 16-bit pro­ success in 1979. We had really made product. VisiCalc.- led to. cessor would help. although "16-bit" it with our floppy disks and VisiCalc. According to any research we could really doesn't mean that much. and it looked like we were going a dig up. many people were buying the Whatever we do will be compatible. long way. So we decided it was time Apple II for small business because because we don't want to alienate our to start putting together a real com­ it had a disk drive and it could run existing customer base. If we can pany. a big company. We needed to VisiCalc. These weren't the people come up with fhe right machine. then start staffing up and hire a lot more was closest to. so I kept my mouthI we can start to talk about some really engineers. So we set the Apple Ill shut because I was only one out of good software and compete well with project in motion. the staff of 15. So we started staffing IBM. Even if we do, the new features The executive staff felt it understood (continued)

THE SLICER 2 additional serial ports for Real 16 Bit Power on a Single Board ­ asynchronous RS232C or Featuring the Intel 80186 111 synchronous communication (Zilog SLICER-THE Complete 8 MHz 16-bit micro- 8530 SCC) processor on a 6" x 12" board Real Time Clock with battery backup 111 for continuous timekeeping 256K RAM, plus up to 64K EPROM • SYSTEM THAT SASi port for hard disk controller Centronfcs type paral lel printer port 11 Two full function RS232C serial ports Fully assembled and tested only $800 111 111 with individually programmed Available in several kit forms also 111 GROWS TO FIT transmission rates -50 to 38.4K baud THE SLICER PC EXPANSION BOARD Software compatibility with the 8086 Gives your Slicer high performance and 8088. video capability YOUR NEEDS 111 8K of EPROM contains drivers for IBM compatible monochrome video peripherals, commands for hardware Video memory provides 8 pages of 111 • checkout and software testing text or special graphics capability 1111 Software supports most types and 2 IBM type card slots for color video, sizes of dis� drives· 1/0 expansion, etc. 111 111 Source for monitor included on disk IBM type keyboard port Bios supports Xebec 1410 and Fully assembled and tested only $600 11 11 Western Digital WD 1002 SHD Available in several kit forms also 111 controller for hard disks Also available: The Slicer 80188 system. Fully assembled and tested only $995 5 1/4" form factor. Call or write for Also avai lable in several kit forms niore information. THE SLICER SYSTEM EXPANSION BOARD Operating systems are CP/M 86 by For expanded memory, additional ports, , Inc. ($85), and MS DOS and real time clock by Corporation ($175). Up to 256K additional dynamic RAM MasterCard, Visa, Check, Money Order, or C.O.D. Allow four weeks for delivery. 2 RS232C asychronous ports 11 Prices subject to change without notice. with baud rates to 38.4K for 1111 serial communication

Slicer Computers, Inc. 2543 Marshall N.E., Minneapolis, MN 5541 8 61 2/788-9481 •St. Te lex 501357 UD SLICER. SLICER

BYTE • JANUARY !985 Inquiry 318 174 TEL PACIS WOZNIAK INTERVIEW COMPATIBLE WI TH THESE SYSTEMS: Computers:

Apple'" lie up and building an organization and emulation mode you only had access Columbia'"II, II +, lie, Compaq',. a management structure based to 48K of memory-you couldn't use Corona'" around doing this new product-the the 80-column display. the extra Heath'" Apple Ill. graphics modes. or the extra memory. IBM'" PC,H-1 PC/XT, 00 PC Jr., The emulation mode wouldn't even PC Portable or AT THE APPLE run much of the existing Apple II Ill Kaypro'" We had some problems getting things business software. and there wasn't Northstar'"II, Horizon/10 Advantage done on time with the Apple Ill much Apple Ill software available. Televideo'" TPC II because of (a) our lack of experience Originally, we planned to deliver 1605, Zenith'" Z-1 Z-151 as a group working together. and (b) four applications with the Apple Ill­ 00, Most other systems running not being able to predict project word processing. a spreadsheet. busi­ MS-DOS '" , Apple-DOS, or CP/M lengths well enough. ness graphics. and a database pro­ We started hiring intermediate levels gram. 's thinking at the time Modems: of managers. Sometimes the man­ was, "People don't really want to buy Anchor'" agers were getting hired at a rate that a computer. They don't want to know Apple'" added a completion date to the pro­ about microprocessors or cards or Bizcomp'" ject faster than it was getting com­ buses. They want to buy VisiCalc­ Bytcom '" pleted. they want to buy a solution:· So we CTS'" Datec '" (Pal Modems) Around this time we started devel­ were going to provide the four major Hayes'" oping a perception of market separa­ solutions. But because we were hav­ lncomm'" tions-good strong separations be­ ing problems managing the Apple Ill lnmac'" tween products so they don't overlap. project while we were building our Multi-Tech '" Yo u don't want to design a product management structure. we were only Novation'" ("SmartCat Plus" that competes heavily with your own able to deliver our operating sys­ Series Only) existing product. I claim that's untrue. tem-SOS-and VisiCalc. which was Prentice'" (Pop-Com) What you really don't want to do is done by Personal Software Jwhich Prometheus'" design a product that doesn't offer later became VisiCorpJ. Quadram'" any more than your existing product. The Apple Ill shipped very late and Qubie'" Racai-Vadic'" (Maxwell Modems) So we started setting up strong had 100 percent hardware failures. Rixon '" (R212A, PC21 2A) boundaries. The Apple Ill would be This is very subjective. and some Transcend'" (PC Modem Card) our 80-column business machine and people might disagree with me. but U.S. Robotics'" . have 90 percent of our market. The I think we were trying to be too pure. Ven-Tel'" Apple II would be our 40-column We wanted to do it on one PC board. This list concerns only home/school machine and have I 0 not two. and it didn't fit on one PC modems AUTO-DIAL. percent of our market. The entire ex­ board. So we got a company that ecutive staff was sure that once the could put three traces between two Apple III was out. the Apple II would IC pins. had them do the PC board, stop selling in six months. I felt really and 100 percent of the Apple Ills down, because this 10 percent were failed. my friends-the hobbyists and home The Apple Ill is really very good. but users. we spent three solid years keeping the The Apple Ill hurt Apple in many Apple II down, and now it's finally be­ ways. but it was a very well conceived ing allowed to grow in that direction. product. And because we were so We've come out with ProDOS. which successful with the Apple II. we de­ is a major improvement. and the Pro­ cided to build in an Apple II emula­ File Jhard diskl is available for the tion mode to take advantage of all the Apple II now. It's a good start. I think software that was out there. The emul­ they are going to find out that allow­ ation mode did get built in. but. ing the Apple II to get there will im­ because of our concept of market prove the whole Apple image. separations. jt was a very limited The Ill will do very well in its estab­ emulation. While our Apple II custom­ lished vertical markets forever. but it 'llllllllllllllllllllllllllllllllllllllll ers were adding 80-column cards and really won't make the huge success Telpac '" is a registered trademark of 16K RAM cards to their machines. we we thought it would. It may have the U.S. Robotics, Inc. actually added chips to the Apple Ill best chance it's ever had. ProDOS is VT is a registered trademark of Digital to prevent access to many of its the best way to get someone closer Equipment Corp. features during emulation mode. In (continued) "Emulates VT'" 52/1 00; Te levideo 91 0; TTY (ASCII) Inquiry JANUARY 1985 • BYTE 3 54 177 WOZN IAK INTERVIEW

more difficult for outside manufac­ done. So we'll give out enough infor­ turers. because they have to supply mation to do this and we won't give 'We did SOS a disk and instructions on how to up­ them any more. because they might date your system. It would have been try to do something they're not sup­ 3 years ago, and the best of both worlds if you could posed to do:· The right way for one plug in a card with a ROM on it and person is not the right way for an­ the rest of the world if the first byte in the ROM were a I 2, other. We closed that machine up to for example. the system would where somebody could have a very recognize that the ROM holds a difficult time finding out how to add hasn't caught up.' device driver and link to it auto­ their own I/0 drivers. We did not make matically. That would have made it it easy for the outside world. We to SOS. We did SOS three years ago, much easier for the card manufac­ thought we wanted all of the markets and the rest of the world hasn't turer. It would have been really easy for ourselves. caught up or come close to it. Macin­ to allow both techniques. but the Ap­ Yo u have to let the end users devel­ tosh went in a different direction, so ple III engineering group didn't want op their own standards. Yo u've got to I can't compare the two systems. to do anything the way it was done give them the freedom to discover on the II. Marketing just didn't have how they're going to use an operating BYTE: Is SOS reall� that good? enough leadership and control. system, what sort of things they're go­ WozNIAK: I think it's the finest oper­ Anyway. having the Apple III as a ing to buy. And if you're really right ating system on any microcomputer failure didn't really hurt the company and have provided a good solution. ever. It's the greatest thing in the much. The Apple II was still very that's where they're going to settle. world. but I wish we gave out listings healthy, but for the next three years The thinking on the III was very much of it. the Apple III hurt the company tre­ like a religion in that it could only be BYTE: Wasn't it the first commercial s�stem mendously because everyone in done one way-our way. We made it that actuall� had installable device drivers? Apple knows how great a machine the very difficult for outside developers, WOZNIAK: Ye s, the Apple II uses Apple III really is. It's a very clean instead of providing all the informa­ device drivers in ROM. on the I/0 machine, it's easy to use. and it's really tion as we did with the Apple II. cards. but they won't work on the III been organized right with the oper­ because they depend on specific ating system. BYTE: Has that attitude changed now? memory locations. On the IlL you Unfortunately, we made it very dif­ WozNIAK: No. It's still the most nega­ load a device driver into RAM for any ficult for anyone to get access to the tive thing in our whole company, and device you add to a system. so it's in­ insides of the machine. We had hired it will be for years. finitely flexible. Yo u can always change some very bright people who figured. I think that when a new market it correct bugs, and it's clean. But it's "This is the right way it should be (continued)

ORDERRECEIVE KEYBOARDNOW & MODULE FREE!

Comput1-'SOerb-53-B SystemsYTEK Corp. f305) 994-3520 4089 S Rogers Circle ·· Te lex: 4310073 MEVBTC BYTEKBoca Raton® FL 33431 Distributor Inquiries Welcome

178 BYTE • JANUARY 1985 Inquiry 44 Inquiry 387

WOZNIAK INTERVIEW

evolves. like personal computers did. schemes for adding up to a megabyte Starting At there's a period of time when you've of memory. we came up with a $695.00 got to let the world go in all random method of adding 64K to an Apple directions. and eventually it will sub­ lie. which was more difficult to use CO-PROCESS! NG side because it wants standardization. and somewhat limited. We refused to The most cost effective way for Z80 Then. once it's obvious what the stan­ acknowledge any of the good 80-col­ system owners to obtain 16/32 bit dards are. they should be heavily sup­ umn cards that were in the outside processing power and software ported by the manufacturer. Yo u can't world-only ours. which had a lot of compatibility is via the HSC C0-16 Attached Resource Processor. try to dictate a standard. problems. At one point during the Apple C0-1 6 is compatible with any Z80 III system running CPM 2.2 or CPM 3. THE APPLE AND THE APPLE development. I wrote some fast disk A few examples include: routines for our Pascal system on the When we cameII out with the Apple IllIII. • KAYPRO 2/4/10 • TRS 2/12/16 II. And I got a lot of flak from Apple • AMPRO LITTLE BOARD the engineering staff canceled every Ill engineers who felt that · they • HEATH 89 • SUPERBRAIN Apple II engineering program that shouldn't go on the Apple II. Nothing • XEROX 820 • TELEVIDEO 802 / 803 was ongoing. in expectation of the • MORROW • EPSON QX-1 0 on the II should be allowed to run as Apple III's success. Every single one • LOBO • OSBORNE 1/EXEC fast or faster than the Apple Ill. That • CROME MCO • Plus many more was canceled. We really perceived was the thinking that stuck with the C0-16 that the Apple II would not last six company for three solid years. months. So the company was almost Every C0-16 is delivered with It was unfortunate the way things all Apple Ill people: and we worked •16/32 bit micro processor •16bit worked out. because we probably put Operating System • 256 Kilo RAM for years after that to try and tell the $100 million in advertising. promo­ • Z80 interface • 16 bit RAM disk world how good the Apple Ill was. driver • CPM80 2.2 RAM disk driver tion. and research and development because we knew. • CPM 2.2 or CPM 3 compatibility into a product that was 3 percent of There is a lot to somebody's per­ • sources with tools • hardware our revenues. In that same time frame. diagrams • board level or case with ception or image of a machine and think what we could have done to im­ power supply. how good it is. How many of my prove the Apple II. or how much C0-1686 friends have them? How many people could have been done by Apple to The only Z80 16 bit co-processor in the world have them? The Apple III give us products in IBM's market. includes • INTEL 8086 • 6Mhz no was a failure the first year as a wait states • MSDOS 2.11 • IBM product-it had a bad image. When BYTE: Are you putting more resources to BIOS emulator • Memory expansion you give a bad first impression. you to 768K • 8087 math co-processor work in that direction now? • 3-channel Real Time Clock • Runs can go for five years trying to over­ WoZNIAK: Ye s. but things don't change many IBM PC applications • Shares come it. in six months. In the Apple Ill we had hard disk space with CPM80 • PC If you looked at our advertising and a beautiful machine. and we spent a diskette compatilibility on many R&D dollars. everything we did here lot of money to try and emphasize systems • CPM86 • Concurrent CPM is coming. was done first on the Ill. if it was that. We were trying to force the world business related. Then maybe we'd C0-1668 to finally accept the machine because consider doing a sub-version on the we knew just how good it was. The only Z80 16/32 bit co-processor II. To make sure there was a good includes • MOTOROLA 68000 The PCjr had a poor initial reception microprocessor • 6 Mhz no wait boundary between the two machines. like the Apple Ill did. It came out in states • CPM68K • Full "C" com­ anything done on the II had to be the wrong month-the month when piler with UNIX V7 library and floats done on a lower level than on the Ill. Macintosh was going to be perceived • Memory expansion to 1.25 million Only now are we discovering that as the leader. The PCjr was perceived bytes • NS16081 math co-processor • Real Time Clock • Complete soft­ good solutions can be implemented as an uninteresting product. ware development environment on the II. They may try for three years to over­ • 100% file compatible with CPM80 The Apple II was kept out of the come its bad first impression. They • OS9/68 UNIX look alike coming business market to keep the III going. may put a lot of their corporate in February. to give our users only one choice. We resources into trying to promote the Hallock Systems Company wanted to make the Ill the success it PCjr and lose sight of the PC. The PC 262 East Main Street hadn't become and that it deserved. will keep selling because it's been ac­ Frankfort, New York 13340 Unfortunately. we made sure that (315) 895-7426 cepted and there are a lot of after­ the Apple II was nowhere close to the market companies out there selling market that the IIBMI PC took. We software and hardware for it. But if made sure the Apple II was not al­ IBM neglects it enough. we may have lowed to have a hard disk or more a chance to turn it over on them in than 128K of memory. At a time when · the next few years if we have a better outside companies had very -usable product. • 180 8 Y T E • JANUARY 1985