The Rise and Fall of Ashton-Tate
Vulcan & dBASE
Cecil Wayne Ratliff (goes by Wayne) was born on the 10th of December in 1946 in Trenton, Ohio (near Cincinnati). In college, Ratliff was designing a small, two seater, rear-engine, sports car, and he was using a CDC 6400 to help himself in the pursuit. He wrote programs to help figure out engine displacement, suspension, center of gravity, and other related problems. To his surprise, he found he enjoyed programming more than he enjoyed automobile design. Before finishing his degree, Ratliff got a job as a computer at Martin Marietta in Denver (before the electronic computer, people who did calculations for a living held the professional title of computer). Calculations would be sent to him, and he’d work them with his computer, and the send back the results.
In 1969, Ratliff was drafted into the US Army to support the US war effort in Vietnam. He wasn’t sent into combat on account to his programming skill, and he was put to work on a logistics war game called LOGEX which was written in COBOL. While he did get to do some programming work, he reports that most of his time was spent ordering equipment and supplies.
After two years in the Army, Ratliff went back to Martin Marietta, and he was a contractor for the JPL. He was a member of the Viking program, and in 1976 he was responsible for creating the MFILE data storage and organization system for the Viking lander support software, a sort of database management system (DBMS). This got him interested in natural language software in general.
That was in 1976, around the time that I became interested in designing and experimenting with natural language, so I bought an IMSAI 8080 8-bit computer kit and put it together. It took a year to put the thing together, mostly waiting for parts. I had to solder more than 2,200 joints. Of course, if I could have bought it assembled for the same price, or even close, I would have. Once I had put it together, all I had was a computer. Nothing was included except 1K of memory. You had to keep buying things, such as a keyboard. I had already spent $1,000 for the kit, then I had to spend another $159 for a keyboard. Eventually I ended up spending about $6,000.
In January of 1978, Ratliff had covered a room with Monday morning newspapers as he was attempting to analyze football pools.
I was in a football pool where you picked the winner and then a point spread. I've never known that much about football. It was the motivation to win, rather than the game itself, that interested me. I thought that if I devoutly applied myself to the mathematical process, I could win.
At that time, Monday morning newspapers would publish the statistics for games that had occurred on the weekend prior. At around 4 weeks into the season, Ratliff had newspapers spread across a room and he was going from paper to paper trying to pull information together. He felt that this was a job fit for a natural-language database system, and he had his IMSAI running PTDOS with which he could do the work. He bought books on artificial intelligence and natural language, and he would study and get pulled one direction or another, and he’d do some experiments. He wasn’t making a ton of progress at first, and then he thought about the work he’d done for Viking. He’d create a DBMS and have the natural language system use it to store/retrieve data. This became useful quite quickly, and just as cars lost his interest to programming, so did natural language systems lose his interest to database systems.
More formative for his new DBMS than MFILE was the JPL DIS database system. He didn’t use any of that system’s code, but he did borrow a few command names like: STORE, DISPLAY, and LIST, and he took some design cues from it as well. Wayne had lost interest in the football pool quickly, and so the first use of his DBMS was in helping him to prepare his taxes. When he had something usable about a year after starting development, Ratliff chose to name his program Vulcan, made a port to CP/M as that was the most popular operating system for the 8080/Z80, and took an ad out in BYTE Magazine.
The ads worked quite well, and sales were good. Unfortunately, Ratliff was burning the candle at both ends, and he was getting a bit burnt out. He was working at JPL during the day, and he was running his one-man business at night. He was handling software changes, responding to mail from BYTE’s readers, taking and making phone calls, and he was physically shipping the product. Using Vulcan, he was able to automate a good bit of the workload, but this wasn’t enough. For example, Ratliff was copying disks himself, one drive to another, and one copy at a time. He was also printing out the manual himself. These tasks took considerable time. He was ready for change, and he felt that he needed to focus on the program and leave everything else to other people. To that end, he had been talking to a professor at the University of Washington about the professor and his wife taking over the marketing duties for Vulcan. From what I can tell, these discussions were happening over many days at least, and during this time period, Hal Lashlee and George Tate called Ratliff about Vulcan. They visited Ratliff at his home, and he gave them a demo of the software. For Ratliff, this was exactly what he’d been hoping for. Here was a company, Ashton-Tate, that was wanting to take over everything he didn’t want to do, and they were just 15 minutes away from him! He signed an exclusivity agreement shortly thereafter.
One very important change was quickly made: the name Vulcan was dropped as there was an operating system by that name made by Harris Computers in Florida. The name dBASE II was suggested, and there was no objection. Around the same time as the name change, Ratliff felt that it was time to abandon accommodations made for teletypes within the interface, and he moved the interface to being completely oriented toward screens. The retail price of dBASE II was $700 (around $2000 in 2023).
Up to this point, Vulcan/dBASE ran on PTDOS and CP/M on the 8080 and Z80 CPUs. It had quickly become a standard of the CP/M software suite: WordStar, VisiCalc, dBASE II. Despite being a standard, it did have its haters. Early software made compromises, imposed limits, and often could be aggravating to use. This was largely due to the memory constraints of 8 bit computers. You can’t exactly have a ton of columns and rows in a database if you’re limited to 64k RAM (and most people didn’t even have that in the early 80s), but naturally people tried anyway, and they’d then complain it was the DBMSs fault. These gripes about column limits (initially 16, later expanding to 32), or speed, or whatever else didn’t stop people from using dBASE. Some complaints were handled by 3rd party application developers, and there were many. dBASE was very often used as the database system for other professional applications, but applications were also written to fill holes perceived by users of dBASE, adding or supplementing functionality.
In 1981, IBM was preparing the PC. Big Blue had the intention of making absolutely certain that all major software titles were available for the PC, and that the PC would thereby fit every use case possible for a computer at the time. Among the applications they wanted available was dBASE II. Ratliff made the port. One year later, Ratliff left Martin Marietta and JPL and he went to work at Ashton-Tate full-time as VP.
The rise of the IBM PC lifted many boats, sank others, and transformed the world in profound ways. Ashton-Tate rose with dBASE II, and dBASE II rose with the IBM PC. Reviews were positive, as the May 1983 BYTE article “A Comparison of Five Database Management Programs” shows:
The dBASE II system can select, edit, manipulate, and display or print any record or groups of records in a file easily and quickly. Simple application programs are very easy to accomplish even for an inexperienced computer user. The dBASE II programming language is somewhat complex and not too easy to learn, but it is very comprehensive and will accomplish almost any programming task.
Later in that same article, Jack L. Abbott says that he’d recommend dBASE II overall. I’ve also heard rival opinions cite dBASE’s programming language as easy, and that it was precisely the easy to learn language that made the program so popular. Personally, the ease or difficulty of any given language is rather a personal thing. For some, a language may be beautiful, easy, and pleasing to use while another might find the same language obtuse, ugly, and enraging. To each his/her own.
Hughes LeBlanc ran La Commande Electronique. This company handled distribution of dBASE II in France. LeBlanc claimed in 1983 that one in ten PCs sold included the sale of dBASE II. Whether or not this was the case, Ashton-Tate survived the transition into the 16 bit era, and it was now one of the names associated with software generally: Microsoft, Lotus, Ashton-Tate.
In February of 1983, Ashton-Tate chose to lean in to the use of dBASE for the building of applications. They released the dBASE II runTime. The runTime is exactly what it sounds like to modern developers’ ears; it was the language and database features without the user interface components. A developer could build upon this, integrate it into a new application he/she was writing, and spend far less money than the normal cost of dBASE II to do it.
In November of 1983, Ashton-Tate had their IPO and raised $14 million (around 42.6 million in 2023). At this point, the company was over 200 employees, pulled in $43 million (~131 million @ 2023) in revenue and $5.3 million (~16 million @ 2023) in profit.
In June of 1984, dBASE III was released. This was a partial rewrite of the program in C. Many articles will say it was a complete rewrite, but this isn’t accurate. Much as the port to the IBM PC used automated translation tools for the bulk of the work, so did this. Code originally written for 8 bit machines was now twice translated and compiled, and still in use. Yes, substantial alterations, fixes, and additions had been made, but the code base was becoming increasingly difficult to maintain (hence the move to C at all).
In the autumn of 1984, Ed Esber became CEO and Wayne Ratliff left Ashton-Tate. dBASE development continued, and the product kept its status despite growing market competition from both clones and alternatives.
By 1987, dBASE III was holding the number four spot for software titles sold on both PC and Macintosh, but things were changing in the world of databases. Particularly, standards like SQL had emerged, and the client server model of database interaction was becoming the norm. Competition had started to outpace dBASE in these realms, and dBASE IV had seen numerous delays. When it was finally released in 1988 for $795 (~2000 @ 2023), dBASE marketshare had fallen to an estimated 63% of the market, and the initial release wasn’t well received. The program had a lot of bugs, and performance wasn’t great. There was a moment of hope in ‘88 when Microsoft wanted to partner with Ashton-Tate on their upcoming SQL Server product. This hope was ultimately crushed as Ashton-Tate’s fortunes continued to fall, and the combined venture came to nothing. One year later, the market share estimate for dBASE fell to 43%. The decline did slow a bit. 1990 also saw the release of dBASE IV Server Edition which worked with MS SQL Server, and it was reviewed quite well. Improvements to speed and reliability were made for dBASE IV (updated version released in 1990), but things weren’t looking great. Layoffs were occurring all over the company and they were losing quite a bit of money quite quickly.
By 1991, dBASE was still holding on to about a third of the database market, and this made them a tempting target for acquisition. The product was clearly good enough for people to continue choosing it, and the company was cheap. Borland decided it was time to buy, from the Los Angeles Times on the 11th of July in 1991:
Ashton-Tate, the personal computer software pioneer that stumbled badly in recent years, will be acquired by fast-growing Borland International in a stock swap valued at $439 million, the two companies announced Wednesday.
The deal, expected to be final within three months, probably will result in further cuts at Ashton-Tate’s Torrance headquarters--now home to less than a third of the company’s 1,600 worldwide work force--as employees are reassigned to Borland facilities in the Silicon Valley or laid off.
For Ashton-Tate, many people claim that the departure of Ratliff was the primary cause of their change in fortunes. Much can be said for singular, talented, and visionary people, and while I admire the talent and achievements of Ratliff, this cannot be the case. Ashton-Tate did well despite Ratliff’s departure. Some people claim that the cause of decline was over-reliance on dBASE as a product. I also don’t believe this to be true.
Ratliff said that more developers should try marketing and selling software. He believed that this gave him an edge over the competition in the early life of dBASE. He understood his customers and their needs. Ashton-Tate released dBASE IV with serious issues and took two years to address those problems while they worked on other products. The company had also grown to a rather larger size. I would wager that they ought to have spent more time and more resources keeping their customers happy. At the same time, Ashton-Tate engaged in a lot of litigation trying to stop clone makers. So, while sales were falling, the company had grown in employee head count, engaged in expensive legal pursuits, neglected their loyal customers… that sounds like plenty of reason for their difficulties without saying their problems were due the loss of an individual or over-reliance on a singular product.
For Borland, dBASE would prove to be a headache. Microsoft Windows 3 was a huge success in the market and people were buying Windows software like crazy. Microsoft Access 1.0 was released on the 13th of November in 1992 for $495 (~1000 @ 2023), and it worked with Windows beautifully. It also happened to have the ability to utilize dBASE III, dBASE IV, and Borland Paradox databases. Many buyers would find that there were promotional prices of $99 at release in 1992. For a product category that had always been extremely expensive, this was quite a deal. Version 1.1 followed early the next year and had bugfixes, improvements, and the integration of BASIC.
Borland didn’t release dBASE 5 for Windows until August in 1994, and at that point Access 2.0 was being released. Ultimately, Microsoft released a better product at a cheaper price, and the user didn’t have to sacrifice existing data. The market chose accordingly.