Since Virtual Development Corporation consistents solely of me, Gary Robinson, at this point, I'll make this page autobiographical.
The services I provide are:
a) Prescience about business opportunities that, in the future, will be commonly recognized as important.
b) Software engineering solutions to meet those opportunities.
The rest of this page will discuss my track record over the years. I'm 67 so it's been a fairly long career so far...
Tracking Cookie
It appears that I invented the tracking cookie in 1995. There is no findable earlier mention of the concept; Google owned my patent, which disclosed it, until that patent expired.
I had noticed that web sites that wanted to be supported by advertising had to actively sell their ad space, which normally required having "space salespeople" dedicated to that purpose. That was expensive and impractical for the vast majority of new sites.
It occurred to me that a better approach would be for ads to be distributed to web sites from central hubs such that any site that wanted to show ads could simply enroll, and that would be that. But such a solution also meant that, if the hubs had information about each person's interests, they could show the ads each individual would be most interested in. And I happened to have technology (see the section on collaborative filtering) that could make the choices about what ads to show... IF it had data about each person's interests.
I came up with the idea of the tracking cookie as the way to gather such data. It was nonobvious at the time because the cookie was only used by the Netscape browser (it had around 70% of the market share, though), and for security, the cookie was designed so that one site couldn't read another site's cookies.
In fact, when I asked about it in a Netscape-related discussion group, I was told: "You could, in theory, if you had source for your server, hack in someway for it to send the cookie somewhere. There is no existing protocol for doing so."
But I thought that all browsers would eventually adopt the Netscape cookie, and I found away around the security, with no need for hacking into the source code of the server.
My 1995 patent[1] which includes the tracking cookie came to be owned by Google until it expired. It disclosed the technical method for using browser cookies for tracking[2], and coined the term "tracking cookie"[3].
It also discussed solutions for the privacy concerns that would emerge over the years[4].
As recently as 2021, Google and Twitter, in a joint legal brief, referred to the tracking cookie as "Robinson's Cookie."[5]
If you're interested in the details of how this invention came to be, see this[6] blog post.
Collaborative Filtering and Recommendation Enginees
I was also a pioneer in the field of recommendation engines. In the 1980s, I created the first recommendation engine using the idea behind modern recommendations, "collaborative filtering,"[7] although that term didn't exist yet. I had it in active commercial use in that decade. For perspective, credit for creating collaborative filtering usually goes to Xerox PARC's Tapestry project in 1992, or sometimes to MIT Media Lab's HOMR a couple of years later. But a few people knew about my work from the '80s, which is why, for example, I got a call out-of-the-blue from Jeff Bezos in the early days of Amazon to discuss it.
I still have the 8086 assembly language I wrote in the 1980s that is almost certainly the first implementation of CF to be used commercially, and could very well be the first implementation of the idea at all; there is no evidence anyone seems to be aware of of another implementation, or even of the idea being discussed, in the '80s or earlier.
When HOMR was released as a MIT Media Lab project in 1994, my own CF-related math had evolved to the point that it was tested to have more accuracy than HOMR's. (Testing involved predicting movie ratings for individual participants in a database of such ratings.)
The 1980s use of CF was in the context of 212-ROMANCE, a voice-response dating service I'd created. It featured recorded "personals ads" people could respond to through voice mail. Because it takes substantial time to listen to recorded ads, I needed a way of deciding which ads each user would be most likely to respond to, so those would be played first. That's where CF was born.
I quickly realized the future CF could have for recommending music, movies, etc. and that's why, by the time HOMR came out, I had better technology than it had.
Prescience About The Magnitude Of A Dating Business Opportunity
This is an instance of being prescient about the future of what superficially might have been considered a small opportunity at the time, but was actually a big one. It's also a story of a business failure and a major learning experience.
In terms of prescience, the vision I'd had was proven to be correct. Out of all the possible applications for the new voice response hardware boards that were starting to come on the market in the early 1980s, the one I picked eventually led to an Inc. Magazine "Entrepreneurs Of The Year" award, lauding it's vision. In terms of business success, it wasn't the best outcome: the chosen Entrepreneurs weren't me. They were two other guys that implemented the same idea, years later.
Above, I mentioned my voice-response dating service, 212-ROMANCE, which I created in the mid-1980s.
My idea was that it should be a service bureau that worked in conjunction with printed "personals ads" such as those that ran in New York Magazine and served publications all over the country (or world) would be a lucrative business.
But when I talked to relevant publications, I met so much resistance to the idea of integrating voice response with their printed offer that I decided to create a standalone, audio-only service. A standalone service was not what I thought was the ideal solution, because it was a lot quicker to scan printed personals ads than listen to recorded ones. But at the time, voice-response technology was so new that the people I spoke to at the publications just couldn't imagine doing what I was suggesting.
It was a classic example of an idea being too early. People couln't accept it yet; it was just too foreign.
My standalone service eventually grew to profitability. Competing standalone services came and went; mine kept growing. Mine, and its competitors, found they could advertise themselves most effectively right on the same pages as personals columns appeared.
But apparently, that process got the publications more and more ready to simply advertise their own voice response systems in the same places as we were advertising. And that created the opportunity for service bureaus to emerge to satisfy that need. As this shift occurred, I had made the mistake of stopping talking to the publications; and by the time I saw what was going on, it was too late. The publications very quickly shifted modes to use voice response; and I couldn't respond to the change in climate fast enough.
I made two huge mistakes which I learned a lot from:
-
I was tricked by the initial resistance of the publications into feeling that they'd never change their minds. Moral of the story: Trust the logic. Don't worry about initial resistance. If the idea is right, the world will eventually come around. You don't have to believe what anyone says. But... you do have to be right. Bob Metcalfe, the inventor of Ethernet, famously said: “The only difference between being a visionary and being stubborn is whether you are right or not.” Only you can take responsibility for being right and trusting that you are.
-
I didn't keep in contact with the publications to see if they'd changed their minds. I should have done so over the years. Moral of the story: While you should have confidence in your logic, you should nevertheless keep doing whatever is required to see whether your assumptions are still valid. In this case, I was assuming that the publications hadn't come to see things differently. They had.
With regard to prescience: I predicted, years before, the specific use of the voice response boards coming out in the early 80s that would win Inc. Magazine's Entrepreneurs Of The Year award with special congratulations for vision, when someone did exactly the same things years later.
Spam Filtering
This isn't a matter of prescience but it seems worth mentioning as engineering.
I created a statistical approach to spam filtering that was used by a number of award-winning spam filters, including SpamBayes, SpamAssassin, and SpamSieve (which won MacWorld's Software Of The Year award across all categories). The Linux Journal article I wrote about it[8] stressed the collaborative effort of the creation of a full implementation and its testing, and the help I received in those areas was most wonderful and appreciated. But the statistical approach was mine.
It improved over the state-of-the-art at the time by being better able to detect when an email had so much inconsistent evidence about how to classify it that it should be marked as "undecided."
One side-effect was that when one of my sons was taking a computer science course at Yale, his professor talked about how SpamAssassin had that feature; that was a bit of fun for my son to hear and for me to hear from him.
That Time Steve Jobs Came To My Blog To Argue With Me About Apple's Direction: In The End Apple Did As I Was Suggesting And That Jobs Had Argued Against
I get a kick out of it, so I'm mentioning it. Not that I think it's of profound import.
In 2004, Apple was selling albums and single tracks. Steve Jobs had been heard to say that "people want to own their music." In general, in the music industry and even in the technical community, the consensus was that music would continue to be sold rather than accessed by a subscription model.
To me, it seemed clear that a subscription model was better and would win in the end. I posted on my blog about how Apple should move to subscriptions[9]. I suggested that perhaps Steve Jobs actually thought that the subscription model would be the way to go in the long run, and was publicly saying the kinds of things he was saying for tactical reasons, but he responded in a comment arguing against me[10].
A few years Apple later added a subscription-based streaming service that worked exactly as I had been advocating. And a bit after that, it stopped selling music entirely.
[1] https://patentimages.storage.googleapis.com/c3/d4/40/239073914fa7fc/US5918014.pdf.
[2] ibid, column 9, line 38
[3] ibid, column 10, line 9
[4] ibid, column 6, line 52
[5] https://s3-us-west-1.amazonaws.com/ptab-filings%2FIPR2021-00485%2F3; see, for example, the top of page 40.
[6] https://www.garyrobinson.net/2021/07/did-i-invent-browser-cookies-for-advertising.html
[7] https://towardsdatascience.com/intro-to-recommender-system-collaborative-filtering-64a238194a26. My original implementation in the 1980s used the nearest neighborhood approach described on that page, albeit with slightly different math. HOMR used that basic approach as well.
[8] https://www.linuxjournal.com/article/6467
[9] https://www.garyrobinson.net/2004/10/steve_jobs_vs_s.html
[10] https://www.garyrobinson.net/2004/10/steve_jobs_vs_s.html?cid=2388224#comment-6a00d8341da47953ef00e550399c508834