Advent of Computing - Episode 153 - The Keypact Mystery

Episode Date: March 17, 2025

When I was down at VCF SoCal I ran into a strange machine: the Keypact Micro-VIP. It's a terminal without a keyboard, covered in dials, with a speaker and a switch labeled "voice". This chance encounter with the unknown sent me down a wild path. It involved the creeping spread of computing, chicken feed, door to door life insurance salesmen, and at least one early hacker.

70% of you will ignore this message about computer-assisted financial planning. 20% of you will read it, but not believe it. The truth is, about one reader in 10 will mail in the coupon and end up spending a profoundly educational and rewarding hour with me and my key-packed computer. The computer is not for sale. It is actually a portable computer terminal that enables you to talk with a much larger computer at no cost. What do you say to a computer? You give it a little information and you get a lot back. For example, let's say you give

the computer three simple numbers. The amount of life insurance you and your wife own, the assets owned by you and your wife, and your liabilities. Within seconds, the computer will respond with figures that compare the tax impact on two estates under three different estate planning options, and you may discover you could be saving up to 50% depending on the size of your estate. The Naples Daily News, Naples Florida, Sunday, May 25th, 1975. Welcome back to AdRent of Computing. I'm your host Sean Haas and this is episode 153, Key Pact. What can I say? I'm excited to be back to my normal schedule.

Now, the genesis of this episode comes from, surprise surprise, my breakdown at VCF SoCal. I love these conventions in part because they are truly inspirational. One of my favorite displays this year was Jerry Ellsworth's collection of obscure computers. It was a whole booth of, mainly, business-oriented machines. That's basically candy to me. On one side of this booth was a machine she had just acquired, the key-packed micro-vip, or micro-VIP. It's ambiguous how one should say that. She said there was almost no information on the machine, and I was hooked. It's not just the obscurity of the micro-vip that drew me in, there are also some peculiarities of this machine that really got me. I like to think

I've built up a pretty good sixth sense for interesting stories, and this machine, well, it triggered a lot of my alarm bells. The MicroVip was a terminal used by insurance salesmen. It was used in the 1970s. It was built into a briefcase. It used an old-style modem with a cradle and everything. But that's where any sense of normalcy ends. And to be fair, this is already on the fringe of what could be considered normal for computing. So, get this. The terminal has no keyboard. It does have a thermal printer to type out data, but no keyboard. Instead, there are two rows of dials. I can't remember the right name for them. They're those kinds of dials that you view edge on and you rotate by moving a little geared handle.

You use the two rows to toggle in numeric data, so, in effect, your only input is two long strings of numbers. The machine also came with overlays that broke those numbers down into fields. That, plus a few switches and buttons, lets you submit numbers back to... well, back to something. Over a phone line. That would be enough of a mystery for me. I mean, come on, it's a portable terminal with a totally unique interface. But there's one other piece that really got me. The keypact could talk. The final output on the device is a speaker.

There's a switch labeled voice that, when flipped, pipes sound directly from the phone handset to the internal speaker. Why is that there? What did the machine say? Therein lies another major mystery. As such, I've decided to bring my powers to bear against this puzzle, and I think I can clear a few things up. I believe I've been able to put together a correct story, and if not, at least it's an interesting one. In this episode, we're tackling the mystery of the key-packed micro-vip. Along the way, I want to look at early portable computing and just how advanced this setup was. This is also the perfect time to examine the brass tacks of how business got automated.

This strange machine should serve as a wonderful gateway into just what computing promised in the 1970s. The computer revolution has fundamentally changed how we live our daily lives. It's easy to discuss how that's occurred. It's easy to look at the history of that revolution in terms of that end goal. Computers got smaller, cheaper, easier to use, interfaces got nicer, the internet developed, that sort of thing. It's often easy to overlook the boring side of the revolution.

It's for that reason that spreadsheets, back offices, and filing systems have always held my fascination. It's the historical path less traveled, so to speak. The boring stuff is a fundamental pillar of computing's promise to humanity. If you can automate away all this work, from accounting to typesetting, document duplication to mathematics, then us humans can focus on more meaningful things. Researchers can do research instead of getting stuck filing papers. We've seen this again and again here on Admin of Computing, from the recent episode on the venerable FlexoRider or VisiCalc, all the way to my coverage of early Hypertext.

The computer is a supreme labor-saving device first and foremost. This was known at least as early as 1946, but probably earlier. Research automation was the first target for computing, but business operations were a close second. As with everything, business computing follows Moore's law. We start with big computers and they get smaller and more capable over time. An office may have a single mainframe

automating a few processes. Maybe payroll gets moved over first, then tax filing and on down the chain. As computer time becomes cheaper and you get more terminals set up, more tasks get digitized. Eventually, the office desk transforms into a stand for your own terminal,

your own window into the computer revolution. What we end up with is an expansion of the horizon of computing. own terminal, your own window into the computer revolution. What we end up with is an expansion of the horizon of computing. That eventually leads outside the office, but how do you bring a computer into the wide open world? In 1969, we don't have microprocessors. We have only limited integration. How do you make computing portable?

Practically speaking, that wasn't possible. A quote-unquote portable computer in 1969 was something that fit in a truck or on a plane. I mean, the Apollo guidance computer was technically portable, but it was moved around by a literal rocket. The solution was to keep the computer stationary and lug around some way to talk to it. In other words, a portable version of a computer terminal. What does this all mean for the keypact? Well, in practical terms, this gives us a time window. We get the first practical portable computer, the Osborne 1, in 1981. At that point, you can already carry around a computer with very few caveats beside, well, the weight.

We have this decade-plus space where portable computing starts to just barely appear, but we don't have proper hardware. One of the first players to make an attempt at this space is called Computone. Through them, we can track a truly wild story! In 1965, William Glover left his job at IBM, and he founded a company called CompuTroll in Atlanta, Georgia. And yes, that's our IBM connection for the episode. Glover had a simple dream. Use centralized computing to modernize the poultry feed industry. And I know, that sounds like a joke. It seems to have been taken that way at the time as well.

In 1967, after Glover picked up some steam, the Northwest Arkansas Times ran a story about CompuTrol. The journalist is, rightly so, a little concerned with the encroachment of digital technology, but he eventually concludes that, quote, nothing, I feel, can be all bad that concerns itself with chicken feed, end quote. As silly as this all sounds, Glover had a very sound core idea. He wanted to, essentially, sell computer time with some extra steps. His business was centered around a single mainframe, an IBM System 360 housed in a computer room in Atlanta.

That computer was hooked up to a phone line. It could be reached remotely by dialing its phone number. The network layer here was plain old copper phone line, as reliable and commodity as it came. What's more, this wasn't even using a leased line or anything special. Phone calls could come in from just a normal phone network. So far, this is reasonable stuff, right? There were a good number of companies that sold computer time and offered remote connections. Usually, a user would have a personal terminal and use that to dial in to a mainframe where they rented time. Some companies

existed purely on this rental model, while others would rent out spare time to recoup costs. This could work as a business model, but it had a very small market. You're basically chasing business from existing practitioners. Your user base are all folk that know how to use a computer. That's very handy, that's very viable, but it's not exactly a revolutionary idea. And in 1960, this would have been a pretty small user base. Sure, it was growing every year, but it's still pretty early on that growth curve. Glover wanted to target non-computer users, with his first target being chicken farmers. Go figure. Now this is part of the whole promise of the digital future thing. A computer can solve

a great many problems. It's just that sometimes you need to go looking for problems yourself. In the field of poultry, there's this thing called least cost feed. Chickens need to eat to grow. The market cost of a bird is directly related to its size and the quality of its meat. It turns out that the quality of the feed has an impact on that. So you can start to extrapolate this relationship between feed quality and profit. Chicken feed is, itself, also governed by market forces. Feed is mixed from multiple different ingredients ranging from chicken byproduct meal to delicious corn gluten flake. The prices for each of those inputs varies. That affects the

final price of the overall feed mix. However, you can't just feed birds on corn by-product alone. For a bird to grow healthy and wealthy, they need a correct mixture of nutrients. This is a classic optimization problem, and if you know optimization problems, they're kind of tricky to work out. To get your maximum profits, you need to make the cheapest feed mix possible while retaining a certain level of nutrition content. That correct mix is called the least cost feed. It could be calculated manually, but that's a tedious project.

A computer could solve the problem in mere seconds. That's what Glover set out to do. But this leads to the classic problem of adoption. How do you get poultry farmers to use a computer? Now this is a tangent, but this reminds me of an absurd IBM commercial from 1977. I know. Weird. Stick with me.

Indulge me. In the mid-1970s, IBM launched the 5100, their first attempt at a personal computer. It was very complex, expensive, of dubious utility, and probably a little too powerful for its own good, but it was an attempt. There's this commercial that shows how the 5100 could be used in any number of real-world applications. One scene follows a cattle rancher who explains how he uses his 5100 to calculate feed mixes for his herd. It cuts to a b-roll of the rancher hammering away at the IBM machine and then another b-roll of him feeding his cows.

Now, I'm not saying a cowboy can't program. I, for one, gladly welcome all practitioners into the fold. If you've ever sworn at a compiler, then we're comrades in arms. That said, I can't imagine very many ranchers knew these arts in 1977, much less 1965. Glover, however, had a way around that problem, and it wasn't teaching cowboys how to program. It was, at once, simple and genius. Custom data terminals. If you wanted to talk to a computer, if you wanted a terminal in 1965, you would have gone out and bought a teletype, or maybe a fluxowriter if you were a little old school. You would have a keyboard and a paper feed. All data would go in as well-formatted text and get hammered out in the same way.

You would have to form literal mystical incantations on paper using a souped-up typewriter. That's not exactly user-friendly, but it is very powerful. It could be used for nearly anything you could ever want to do with a computer. But these poultry farmers, they just wanted to do one thing. They only wanted to calculate least-cost feed mixes, at least at first. For that, you don't really need a flexible solution. Glover would design a purpose-built terminal just for chicken feed. It was called the Computrol 1200.

We don't have manuals or schematics or anything very specific, but we do have a pile of news coverage. For a surprising number of communities, the Computrol 1200 would be their first glimpse of the digital future. Now, I do mean that very literally. This isn't a case of me adding some flair or drama to a story. I don't need to this time.

To quote from that very Northwest Arkansas Times article again, quote, For the most part, Fayetteville has had the enduring quality of resisting fads, fancies, and the cybernetics of our advancing space age. Here we are rarely jolted from the comforts of traditional conformity by such mod innovations as miniskirts, topless waitresses, race riots, belted jackets, breakfast martinis, and or computers. Computers, however, more than most modern paraphernalia, have wormed their way into the life of the hamlet. The university performs myriad miracles of efficiency with its machine, and the First National Bank now handles a great variety of business functions with its model.

In spite of their proliferation, they remain totally mysterious to me." This is the mid-60s we're talking about. We focus on stories about computer people here on the show. It can be easy to forget that those folks were few and far between. Most people in the 60s had never seen a computer except maybe in newspapers or sci-fi movies. ENIAC had been disclosed to the public in 1946, just 20 years ago at that point. It sounds silly to us to compare a computer to, well, a miniskirt, but in this period both items would

have been novel and more than a little scandalous. The Computrol 1200 also looked the part. As with all good computers of the era, the Computrol was shaped like a desk. The terminal had no keyboard, which makes it stand out from the pack. Instead, you had a panel covered in hundreds of dials. These were broken up into different input fields for the price of different feed ingredients, and probably for a few other parameters. I say probably again, because no documentation. It sounds like users would dial in the local prices of different feed ingredients on that board, and I believe the nutrition content they were going for.

Now, if this is really the case, and I think it is, then this is a very smart interface. Dials can function like memory. Once you set a dial, it stays set until someone or something physically moves it. By using dials for inputs, the user didn't need to input any values unless there was a change. The price of bone meal may fluctuate a lot day by day, but corn gluten stays the same price for weeks at a time. It would have been annoying if you had to hammer in the price of corn gluten even if

it remained unchanged. This is also a very understandable input device. You go over to a field and physically dial in a number. The number is displayed on the dials. You can check that the input is correct by, well, reading the number across the dials. That's simple, direct, and gives instant feedback. But that's just part of the terminal. Atop the desk was a teleprinter of some sort.

I have no idea which particular teleprinter. The images that I have aren't clear enough to tell, but it has similar stylings to IBM models. This was the output. After you sent in data, the machine would respond, and that response would be printed out for you. The design space there is pretty restricted by the technology of the time, but a printer would have been somewhat recognizable. Teleprinters in this period were basically electric typewriters with the keyboard cut off. Everyone knew how to use a typewriter, or at least what they were.

So that's not really a new technology. It's something that you see and you get. The final piece was the network connection. This was handled by a telephone. And I'm not entirely sure how it was set up. My initial guess was an acoustic coupler, since that would be correct tech for the period, but I'm not too sure.

In the handful of photos I've found, you can see a custom looking telephone sitting on the 1200's desktop. I've even seen one with that custom phone, and a normal bell phone sat on the desk, both with receivers in their cradle. An acoustic coupler is an older style modem hookup. It's a special cradle that you stick a telephone handset into, it has very distinctive rubber cups that grab and hold fast on the speaker and microphone side

of the handset. Then a matching set of speaker and mic in the coupler are used to send and receive audio signals. This was done because Bell Telephone limited what could be directly wired into the phone grid. The acoustic coupler went away as rules about telephone hookups were relaxed. The custom phone looks to have rubber cups but I'm not entirely sure. Usually when a coupler is at play, you'll see a normal phone sitting next to the modem, but the photos don't show that.

So maybe the 1200 did have a direct line and the phone was mounted on the desk as a pass through so you could still make calls. I'm just not sure. But whatever the case, all communication did eventually go over a normal audio telephone line back to the data center in Atlanta.

From all accounts, the chicken business worked pretty well. By 67, CompuTrol has a good number of installations, enough that they have to upgrade their mainframe. They're also hiring programmers and analysts, beefing up the company considerably. CompuTrol also expanded their offerings. By 1969, they started offering a price database. This allowed customers to send in the prices they received for their chickens. Those prices could be analyzed and used to calculate a fair market price. The system also expanded to running payroll for chicken producers.

But Computrol wouldn't stay with chicken feed exclusively. In this same period, they brought on another ex-IBMer, William Robeson. He would take the company in a new and wild direction. The idea, once again, was simple and really an easy expansion of where Glover started. They had a way to do centralized calculations from remote terminals. The idea, once again, was simple, and really an easy expansion of where Glover started. They had a way to do centralized calculations from remote terminals. They had an interface that was proven to work with non-computer folk. Robeson figured that the same core concept could pivot to dominate the life insurance

industry. In 1969, the idea starts to bear fruit. CompuTrol would also change its name in that year to CompuTone. In 1985, Ronald Kessler, an investigative journalist, published a book that didn't exactly set the world on fire. It was titled The Life Insurance Game. One review by Terry Mapes claimed the book, quote, "...might actually have been better as a series of newspaper articles.

After some eye-opening early chapters, it gets bogged down in unnecessary details about computers, insurance investigators, insurance company investigations, the history of life insurance, and regulation of the industry." To me, that sounds like a good time. What can I say? There's something wrong with me. Now, why did this book come to my attention? This takes us down a weird path.

Kessler, as I mentioned, is an investigative journalist. This book was the result of a long-term investigation into the life insurance industry. He was helped along by a man named Ray Thillebert, who, in 1985, was the senior vice president of Computone Systems, Inc. Who is Ray and what is he doing to our beloved corporation? Well, it's actually fully explained inside the life insurance game. From which I will be paraphrasing quite heavily. Copputone was heavily involved in the investigation that led to this book, and Filibert plays a crucial role in how Copputone became an insurance concern. The bridge starts in 1969, as I said,

when Robeson figures the same infrastructure that was used for chicken feed could be adapted to insurance. The issue was that Robeson wasn't an insurance man. He was an IBM man. So he put on his best blue suit and went out networking, and he ended up in contact with Filibert. At the time, Filibert was working for a consulting company called Bowles, Andrews, and Towne, Inc. They were established in 1946 as a general legal consultancy.

There's not a lot of interesting, or even publicly accessible, data about this company. They existed, they have a legal paper trail, they had clients, and they had some concern with life insurance. Kessler tells us, probably directly via Filibert, that BAT Incorporated had been using index cards to manage policies. That data would be used to create policy quotes.

These are, at least in old school terms, sometimes called illustrations. This was done manually, maybe with the aid of a desktop calculator. It was a long and annoying process. And keep in mind, for these types of clerical companies, time directly translates into money. At this point, computing was still firmly the future, and historically, insurance agencies had kept everything as hard copy. But we were on the cusp of a change. From Kessler, quote, Robeson met with Filibert in his office and suggested using computers to perform the analyses. At first, Filibert considered the idea a gimmick, but he woke up at 2am and began planning the

system. By 5am, he had designed six programs, including several still being used by Coputone, to analyze customers' needs and compare policy prices and values. When Robeson arrived at his office the next morning at 7.30, Filibert was waiting there for him with a proposal." This is classic, right? Filibert got bit by the computing bug. The dastardly digital demon slithered its way

into his mind just as it crawled into Fayetteville, Arkansas. For me, this is one of the wild repeated themes in computer history. It's this kind of instant conversion. Maybe some people are predisposed, primed in some way for a nearly spiritual experience. Maybe it's just a feature of this technology in general.

Filibert was given this idea one night and by the next morning, he was a convert. Coputone and BAT formed a joint venture. They worked together to adapt Coputone's technology to handle the kind of insurance tasks BAT was involved with, specifically to generate policy illustrations automatically using a computer. That involved almost the same software as CopyTone's poultry network. There would be a centralized database with rates and policy information, a way to send parameters and requests, then a way to grab and display a result.

There are two key differences here. One is complexity. Life insurance is actually a good deal more complex than chicken feed. That's fine though. Software can scale up and down as needed, and Copytone was hiring a lot of new COBOL programmers at the time. That's not a joke either. Once again, no need for drama, we have the job listings and sources to prove it. There was a major difference though, and that was the user base. These terminals will be put in the hands of life insurance agents.

The interface would need to be easy to use. Once again, I'm not saying life insurance salesmen can't program, but I doubt many knew the arts in this era. There was also a difference in user pattern here. poultry farms are stationary affairs. You aren't likely to up and move your feed mill very often. Life insurance salesmen are, however, pretty mobile. They'll have some home office, but many make house calls. Some travel and ramble across the country. The classic Copputrol 1200 was easy to use, but it was the size of a desk. The new user base required a new type of terminal.

In 1969, ads are taken out in a number of papers. They show a briefcase with four simple words superimposed. KeyPact is almost here. word superimposed. Keypact is almost here. This is where we reach the core of the mystery and where the trail becomes all the more confusing. The initial sale model here is a bit of a maze. Here's my best attempt at untangling it. Computone manufactured these new terminals. There were a number of models all under the keypacked name. They sold those physical terminals to BAT. Then BAT turned around and sold units to insurance agents. Those units were backed up with a subscription to Coputone's mainframe service.

It sounds like it was a pay-per-minute kind of thing. The confusion here is that early sourcing around 1969 and 70 says that KeyPact was manufactured and sold by BAT Inc. There's no mention of Computone since only insurance agents would be vaguely aware that that company existed. There's another layer here as well. Insurance agents, customers of BAT and Computone, would take out ads for their service. Those ads don't always mention the parent companies, just the fact that Joe Insurance from Chattanooga has

harnessed the power of the keypad computer. He's gonna give you life insurance faster and cheaper than ever before. These are paired with local interest stories about how some small-town insurance man has entered the digital age with his key packed, and you very well may see a computer in your own living room very soon. Then in 1970, BAT went bankrupt. Coputone buys up the remnants of BAT, staff included. That's how Filibert becomes part of Computone itself.

It also means that early press starts outsiding BAT, then moves to this weird period where both companies are mentioned, then switches to just Computone. So did we make it? Is this the same terminal I saw at VCF? Well, not quite. I guess it's time we talked about the hardware side of things, and the software that backed it. Once again, we don't have great documentation. What I have to go off of are scattered docs, newspapers, and magazine articles. Luckily, there was a lot of press

from many different sources all covering the keypact. Most importantly, a lot of newspaper photos. In 1970, the first keypact terminal hits the streets. They're built into Samsonite briefcases. From looking at the later model at VCF, it's clear that these were serially produced. Computone wasn't exactly churning out millions of units. In fact, we have hard numbers. In 1970, just before bankruptcy, BAT purchased 1,475 terminals from Computone. Not exactly small numbers, but we aren't to the point where economies of scale really matter. There were a few early models kicking around.

One Computer World article from the end of 69 shows a particularly strange one. It uses the same type of interface as the old chicken feed machine, just scaled down. You get four rows of dials and some switches and buttons for sending data and changing communication modes. Above those dials is an acoustic coupler, and this time I do know it's an honest-to-goodness coupler. You would have connected it by physically picking up a foam handset and smashing it into the big rubber cups. On this early model, there was a faceplate that marked out fields on the dials. This would have worked like the CompTrol 1200, just tiny, that's about it. From what I gather, data would have been sent over the line as a long string of numbers.

On the terminal side, the labels just help the operator properly pack together input data. And on the server side over in Atlanta, well, that would already know how to unpack everything properly. As for this really early model, there's a bit of a hitch. I only have one image, and there isn't any visible form of output. You just have dials and switches and a big, bold acoustic coupler. This makes me think that perhaps this model is a fraud. It's likely some kind of mock-up, which isn't really uncommon. Many companies showed off fake hardware designs before the machines were finalized. The photo of the Altair 8800 that ran in popular electronics was, after all, a shoebox.

In 1970, we start to get all those local interest articles with nice big photos. This shows a new model of keypad. It has four rows of dials and a coupler, plus a single output device. Now, dear listener, I invite you to take a wild guess at the mechanism here. Was it a few digital displays? A tiny LCD? A printer? Computone didn't go for anything as pedestrian as that.

Instead, it's a speaker. A speaker! This is one of those twists that I never expected The later model I encountered had both a speaker and a printer I assumed that you know the speaker would be the add-on But that's not the case the speaker came first That leads to the next obvious question What was being played over that speaker? You see, the keypad terminal

responded via voice, specifically a computerized female voice. That's pretty interesting and pretty surprising. From a purely design perspective, this is great, right? You have all the benefits of dials, and then the output is clear spoken English. That's cool, that's futuristic, that's streamlined, it's easy to use. But from a technical standpoint, what in the world is this? 1970 is very, very early for full speech synthesis. We get the first reasonable synthesis out of Bell Labs in the late 60s, but that's in research.

If Computone was doing full speech synthesis in 1970, that's cutting edge stuff. This is also where my meager sources run a little thin. It's implied that this isn't full speech synthesis. The keypact only reads out numbers. That makes things easier, at least in theory. Some articles also claim that the voice can point out errors in data entry. If true, that would further complicate things. My guess, my complete shot in the dark, is

that Computone used some kind of pre-recorded messaging here. Maybe they had a set of tape loops and some circuits to switch between them. We don't have anything internal, so we can't be too sure. Something we have more information on are inputs. The keypact system was far more complex than anything Compute-Own had done before. There were multiple different programs to run, and a large data set to reference. How do you navigate all that without a screen and keyboard? The solution is delightfully simple. Each terminal came with a set of form overlays. These were sheets of laminated paper that fit over the rows of dials. Each sheet indicated which dial formed which input to the program.

This is one of the reasons that I think the 1969 image is fake. The form fields in that photo are directly on the unit. In production models, there are alignment pegs for these overlays. That's missing from the 69 image. These overlays made it possible for one keypact to pack data from multiple different server programs. It also let you switch between different programs. And this was all a little tricky in its implementation.

Each program on Computone's mainframe had a three-digit ID number. Program 092, for instance, was for retirement planning, while 095 was for calculating tax on investments. The first field on every overlay was a three-digit program number. When you grab an overlay, it says, in big font, what the program is for. It marks out the first three dials as the program number and has the number for that program right above those dials, and again, big bold font. To switch programs, you swap the overlay and then dial in those bold numbers.

This is such a neat way to do this! The mainframe just sees a long stream of numbers. The first three are always the program number, which not only tells the mainframe what to run, but how to unpack the data. It also makes the system very extensible. Nothing is stored on the user side, so you're free to add new programs whenever you want. Updates could be shipped out in a manila envelope, and would only consist of a sheet of paper with some nice holes cut in it.

But the whole voice thing introduced a bit of a wrinkle to the operation. The total workflow went something like this. You turn on the keypact, pick out the program overlay, and fill out the form. You pick up a nearby phone, put the receiver in the keypact, and then dial a 1-800 number. That connects you up to Atlanta, and you're ready to roll. With the press of a button, your data is transmitted.

A few seconds later, a female voice comes out over the speaker. The computer reads out some numbers, and you're expected to write down whatever the computer says on a piece of paper. See the issue? The whole system is pretty foolproof, except that last step. I have no recording of the keypack to go off of, but I can't imagine the voice was all that clear. It could be misheard, or you might just write down a wrong number, maybe miss something altogether. What happens then? If you notice the error, you could retransmit and rerun the program, but that would cost you. You were charged by the minute. If you didn't catch the error, well, that would lead to its own problems. This issue was addressed in 1977, when Copytone released a new model, the Micro VIP, or as

I like to call it, the Micro VIP. This is where we reached the end game of the product, its final form, so to speak. There are at least three models of the MicroVIP that show up over the course of the 70s, but they all follow the same layout. There are two big changes to discuss. The first is the addition of a thermal printer. This is actually something of a return to form, since the old copy-troll machines had used printers after all. Now data could be printed out as hard copy. That was much less error-prone and would allow for

things like printing of complex forms and physical contracts. The speaker for voice outputs was retained. You know, for the cool factor. The other big change was done to make space for that printer. Two rows of dials were dropped. This left the micro-vip with just two rows of inputs. However, that doesn't represent a loss of data. Instead, there are actually more dials than ever. Everything was shifted so the rows are longer. Plus, it looks like they were using smaller dials.

Upgrades are always good and all, but that introduces a problem of compatibility. From what I've been able to piece together, micro VIP programs would have been incompatible with older keypact programs. The data size is just different, plus the whole printer thing. To mark the difference, the program ID field was changed. Now, this is speculation, but I think this is warranted and correct. Some micro-VIP overlays show a four-digit program ID. Some use the older three-digit notation. My guess is that three-digit programs are the older voice-only ones, and the 4-digit ones

would have used the printer and only worked on a micro-VIP. That however, is just my guess from... weeks and weeks of looking at news articles and photos of these things. Now that gets us up to the mystery of the machine at VCF, but there's still more. Come with me as we delve into the true depths of the Computone mystery. There are two final stories I want to share that, for me, flesh things out and really lead to some weird places. The first gets into how Computone made its money and how they became an entire

economy unto themselves. This comes primarily from the life insurance game and a little extra information I got from some news articles. There were some obvious cash flows. Computon sold their terminals for between $1,500 and $4,000, roughly. Numbers change over the years and there's a few different models that aren't really itemized anywhere. Computone also made money off system usage. From what I can tell, it was charged as a per minute rate.

Many articles said something like, it costs as little as a dollar to get an insurance quote, but I think there's a flare and conversion going on there. Computone also sold computer time. This was unrelated to the whole insurance business. Many companies would do this on the side as a way to recoup a little money. We know Computone was running a System 360, and those weren't exactly cheap machines,

so extra income would have been a plus, especially early on. Now there is another rev stream that bears mentioning. This particular feature of keypact is explained in the life insurance game. One of Copytone's big selling points was that it had a centralized database of thousands of insurance policies. That information was sent in by the insurance companies themselves and cross-checked before entry. Insurance agencies actually paid Computone to store and serve this data.

But there's a catch. As explained by Kessler, insurers don't want competitors getting access to their precious rate sheets. This includes agents working with competing insurers. A universal life agent isn't allowed to see rates that Aetna sells at. Computone developed some type of authentication system to prevent this.

I believe this is also how they calculated billing for users. The mechanism uses some sort of access code. Each terminal would send in their access code, which would then be used to determine what that particular terminal was allowed to access. I have no idea how this actually worked. I suspect that each keypact unit had some baked-in ID code, but I have not been able to verify that. Those are the direct ways keypact made money, but there's more to the story. There were groups that offered paid classes on how to use keypact terminals. These show up in a number of places, including Purdue University.

You gotta admit, there's something funny about how folk figured they could offer classes on how to use a device that was designed to be user-friendly. Copytone also spread out into other fields. This was already a proven tactic. The basic technology was sound enough, so why stay in just insurance and chicken feed? The first venture was a service called TelStat, which used the same hardware to provide access to stock market information. In fact, this may give us a glimpse into the server-side hardware. TelStat could be operated from a normal touchtone telephone. The computer took inputs as button presses

and responded via audio. That means ComputeTone must have had some kind of PBX setup for handling the phone lines. I don't know my PBX history very well, and I'm sorry for that. I'll get into it one day. But that should serve as a lead for anyone who wants to chase down exactly how Computone's voice was generated. Telstat itself represents a whole other rabbit hole. It seems like it starts with Computone and then spins off into its own business with its own hardware.

Alright, so one final swerve to hit the final part of the story I want to discuss. Security. This is something that I sort of stumbled into while preparing this episode. Computone may have been one of the earliest victims of an honest-to-goodness hacker. Here, I don't mean hacker as in the old sense of the term. I mean someone may have tried to break into Coputone's servers for pure financial gain. At least, maybe. This first came to my attention when I read a 1972 article in the Atlanta Journal. The headline was, Midnight Rider Seeking to Rob Atlanta Computers. It opens like this, quote, Behold, the computer bandit.

Computer experts say it will happen, and computer companies are thinking of ways to prevent it. But for one Atlanta company, computer cheating is already a reality. The Atlanta company, Copytone Systems Inc., says it has a midnight writer who is trying to steal information stored in the company's computers." And let's be frank here, I love this. Early computer crime is, to me, one of the most interesting things in the world. 1972 is so early in the timeline that we don't have legislation, much less agreed upon language for this kind of stuff. Best of all, the story sounds so familiar.

So here's the details. Sometime in 1972, Compute Ones staff starts to notice something funny going on. Each keypact user had a security code that was used to connect to the mainframe. But someone kept trying to get in without a code. It must have looked like someone was prodding at their machine. This was happening at odd hours, usually at night when the system wasn't at peak use. Someone was trying to break in over the phone line. Robeson suspected that the perp was misusing a keypact terminal.

And that could be possible. I actually found a listing in a newspaper around the same time from someone who lost their terminal. This listing was actually offering a cash reward for any help finding their missing keypact. And that would make for a great story, but I don't think it's the case. I'm not sure exactly how security codes were handled, but my working theory is they were hard coded

to the terminal. At least I've never ran into anything that said you had to punch in a code at any point. My pet theory is that the quote-unquote midnight rider was using some other type of terminal, or maybe a modem hooked up to a smaller computer. I think that's more likely because the alternative is a little less sensical. If you physically had a keypack terminal, you could just get in. If the security code had to be entered on physical dials, well, then it's not very secure. It would just be on the face of the keypact, bright as day.

If you had to dial it in on a phone, well, I'd bet you dollars to donuts that the insurance agent that used the terminal had written it down somewhere, that the insurance agent that used the terminal had written it down somewhere, and probably kept it in their briefcase. And the keypact was built into a briefcase. Tony Amorose, a manager at Computone, headed up the effort to capture the would-be thief. And this is where we get my favorite part from the article. Quote, the thief is a brazen sort of fellow, said Amarrose. When we know he is trying to use our computers, we shut them off, and he phones us and asks, Why in the hell did you do that?

End quote. That makes this into one of those classic tales of folklore. Some computer nerd is out there trying to break into your mainframe over a phone line, no less. When you make a move against him, he calls, not to gloat, but to complain. I've only found two articles that mention this event. As far as I can tell, the hacker's never caught, and the story was never picked up by a larger news outlet. I want to underline how early this is.

This is a textbook case of hacking. It's an outside party attempting to gain unauthorized access to a computer system. When it comes to stuff like this, my standard source is Computer Insecurity. It's a book of case studies published in 83. The book presents cases of computer disasters and crimes. It's one of my favorite texts, since it organizes everything into multiple tables and even provides

citations. Leafing through, I can only find maybe two of these classic intrusions prior to the Computon attack. That's it. That's how early this is. Most other disasters are either embezzlement from the inside of a company, theft of software, or actual disasters like fires, bombs, and airplane crashes. That's not to say the computer crime was unheard of.

It's just that external intrusion was rare. At least, that's one possibility. You see, running into this article has got my mind a little twisted up. I've started to believe that it's less likely intrusion was rare. Instead, it's possible it wasn't covered in the press. Hear me out as this is my last pet theory of the episode. We don't get nice wording and legal frameworks about computer crime until the middle of the 1980s. The term hacking doesn't enter common parlance until the 90s or so.

How then do you even discuss this kind of activity in a popular press paper. The articles that I can find about computer incursions in the 70s are bizarre reads. That is, even when you can find them. In the case of Computone, the hacker is called a midnight rider, bandit, thief, cheater, and eavesdropper. thief, cheater, and eavesdropper. This bandit was trying to quote unquote steal information, but this isn't a spycraft thing. This is about life insurance rate cards. I mean, realistically, this is probably someone having a bit of fun trying to break into a system. Just the lack of common language here, though, makes it really hard to track down similar sources. So it's very possible that there are many more cases of hacking in the 70s that are hidden in plain sight. I would like to personally challenge anyone listening to find me some articles.

This is, perhaps, a huge story just waiting to be untangled. Alright thus ends the mystery of the key-packed Micro VIP. It turns out to be a very custom terminal made to sell life insurance door-to-door across America. It grew out of technology used for chicken feed formulation, and its manufacturers may have been a victim of one of the first hackers. All in all, pretty wild story. Questions persist, but I believe I've managed to solve the majority of the mystery here.

The keypact saga fits pretty well into the larger arc of the computer. When we talk about computers entering the office and how digital technology automated mundane work, well, this is it. This is ground zero for that phenomenon. The keypact as a system automated away hours of tedious calculations and it did so in mass. Its portability and simplicity made that possible. An insurance agent wasn't going to spin up their own program to illustrate plans, but they might buy into some hardware that could solve that problem for them.

It's also wild to me how early the keypact is. Portable computers don't come into the picture for another decade. We don't really have a concept of user-friendliness at this point either. Computone is able to bring both of these ideas to market. And what exactly do they do with all that cunning and craft? Why, they change the way life insurance is sold. It sounds almost too silly to be real. Almost. Thanks for listening to Admin of Computing. I'll be back again in two weeks. I'm back on my normal schedule with another part of Computing's past. If you like the show, please take a minute to share the podcast with a friend. Word of Mouth really helps the show grow.

You can support the show directly by purchasing Advent of Computing merch or signing up to be a patron on Patreon. Patrons get early access to most episodes, polls for the direction of the show, and bonus shows. You can find links to everything on my website, AdventOfComputing.com. And until next time, have a great rest of your day.