They say that 1968 was the key year for breakthroughs in computer engineering. The mouse, graphics terminals, integrated circuits, and AI all have their roots in that period. But from the first demonstrations of breakthroughs it was a long way to make them truly useful.
My first computer was a PDP 8/F, located in the Chemistry department of our local university. I was taking an Advanced Placement class in Organic Chemistry at high school. My Chemistry teacher arranged for me to go to the university once a week to work with a college professor, who showed me how to use the computer.
There were 16 binary switches that were used to set the initial address of boot-loader that would startup the system. This boot-loader would read the operating system from a magnetic tape drive. A teletype was used to enter programs written in BASIC. The program would be written from the memory to a paper-tape punch for permanent storage. I still have a few paper tapes of programs that I wrote in high school.
It was unthinkable at that time that average citizens would own computers. Large timeshare computers were equipped with: batch cards readers, teletypes, graphics terminals, tape drives, and line printers. They existed in data centers with high-powered air conditioning to draw off the heat. They were cared for by technicians in white lab coats.
Programmers used teletype terminals or card punches to create programs. These were submitted to the timeshare computer by the technicians and a stack of fan-folder striped paper was returned the the results of the program output. During college I first started using video terminals (TV with a keyboard attached) to replace the teletypes.
The early 80s saw a large number of small computer introductions for hobbyists. The three main computers that were available when I was in college were the Commodore 64, Apple II, TRS-80. Each of these had a full address space of 64K bytes of RAM available. Byte Magazine published updated specs for all of the computers monthly to make us salivate and dream of what was coming out.
The IBM PC was introduced in 1981 and immediate set the standard by which all other computers were compared. There were several different operating systems that were run on the hardware. Anyone using a computer at this time was clearly a hobbyist who enjoyed computers for the novelty. They didn't actually do anything useful for real people but were an expensive toy.
I started my dream job in 1981, two weeks after I graduated from college. We moved from Arizona to Boise, Idaho where I began working with Hewlett-Packard. I was an Electrical Engineer but really had a passion for programming. I worked for Disc Memory Division of HP where we built a massive disk drive with 7 magnetic platters. I was now playing the big boys on the cutting edge of technology.
Most of the early computers had a lot of limitations and required constant fiddling to get them set up to do anything useful. They were very expandable and flexible but also unreliable. Mere mortals could not do much because there was so little software.
In 1984 Apple released the Macintosh computer which was completely unlike anything on the market. It was a completely integrated system with hardware, operating system software, and application software. It was built into a toaster box with a 6-inch display. It had bitmapped graphics and a mouse unlike anything I had ever seen.
It was so different that I was initially unsure about whether it would do real computer work. MacPaint, MacDraw, and MacWrite software allow anyone to build text and graphics documents instantly. The mouse was strangely useful when working with the graphic window interface.
After the initial confusion wore off I became convinced that the future computers would look more like the Mac than they would like the IBM PC. This has indeed proven correct.
Despite the amazing functionality of the Mac, Microsoft captured the dominate market position with DOS. Compared to the Mac OS, Microsoft DOS was very limited, ugly and hard to use. But Microsoft had chosen a business strategy of running on everyone's hardware. By the mid-90s there were hundreds of computer manufactures that were all producing IBM PC clones, all running MS-DOS.
The market had chosen to buy the cheap hardware and this allowed software developers to build applications for all of the market by making software compatible with MS-DOS. Microsoft knew that they needed to upgrade the operating system with better usability but it took many years.
Apple had adopted for a closed system from the beginning to control the user experience. This was demanded by Steve Jobs who was in full control of the development decisions. This resulted in a great user experience but high prices led to limited market share.
The late-80s were a huge mess as Microsoft attempted to implement a full windowing OS. They worked to copy the Mac and several lawsuits resulted from the melee. Meanwhile customer were stuck using MS-DOS to do their work.
During this time I began working on early imaging software. This provided an interesting vantage point for the Window/Mac debate. Computers were looking for the "killer app", a purpose for a task that they could do far better than any other tools available. This would justify the purchase and training required. Word processing and spreadsheets were the killer apps of the day.
As computers became more capable of processing image data and fonts it became practical to use them for publishing tasks previously requiring a typesetter. Suddenly there was tremendous interest in publishing done by an individual using a personal computer. The Macintosh computer showed what was possible but it was Aldus Page Maker that brought Desktop Publishing to the masses.
To do a great job with Desktop Publishing, it is necessary to excel at Fonts and Images. HP had begun work on scanners to exact images from paper. We were developing both the hardware and software for image scanners.
By the mid-eighties computers were not powerful enough to process images. Opening a single image file could take minutes to display. Both the hardware and software were to develop very rapidly during this time. By the late eighties HP was investing heavily in this product area in anticipation of the coming boom in computer imaging applications.
We had multiple lab teams working on scanner hardware and software. Creating imaging software during this era was quite problematic. The Macintosh platform was built to work with images, so it was very natural to implement the scanner software for the Apple computers. IBM PC Clones were running MS-DOS and were completely ill-equipped to work with images.
An early decision was made at HP to implement the scanner software for windows and bypass MS-DOS entirely. The problem was that only a small fraction of IBM PC users were running MS Windows. Our first scanner application required MS-DOS users to start up Windows from MS-DOS and then load the application software to do a scan. Then they would shutdown Windows in order to do the rest of their work.
At one point we shipped a version of Microsoft Windows on 20 floppy disks so that users could start up Windows to run our software. Looking back this is utter madness. I guess that is the penalty for building products that are lacking key parts of the infrastructure to succeed. I learned valuable lessons about technology timeliness during this era.
Starting in 1988 HP began to invest in OCR (Optical Character Recognition). This is software that allows a scanned page to be converted into editable text documents. The data can be saved in a variety of document formats, such as MS Word and Aldus Pagemaker.
I worked for three years on OCR. Our mission was to create the most accurate OCR for badly degraded text. Our goal was to be ten times more accurate than our nearest competitor on 5th generation photo copies. We met our technical goal and produced the worlds best OCR.
But we soon found that the business goals related to creating a new product line were much more difficult to solve. HP was requiring us to produce a business with $50,000,000 in revenue every year to pay for the software development. This proved to be impossible, so the project was abandoned.
HP could not even find a way to sell the technology wholesale so it sat in the Archives next to the Lost Arc of the Covenant.
Our OCR product, named Tesseract, was originally developed at Hewlett-Packard Laboratories in Bristol,England and at HP in Greeley Colorado between 1985 and 1994, with some more changes made in 1996 to port to Windows, and convert the code to C++ in 1998. In 2005 Tesseract was open sourced by HP. After being abandoned for 14 years. From 2006 until November 2018 it was developed by Google.
I was at a Python developers conference in 2013 where I ran across a seminar on how to use Tesseract to scan receipts. I attended the seminar out of curiosity to see if they were using our code. I was delighted to find that Tesseract had indeed come back to life. 22 years after its death it is still alive and kicking. Google "Tesseract OCR" to find out more.