Vannevar Bush, Doug Engelbart & Virtual Assistants
In the 1930s, Bush had an idea called Memex, a knowledge storage system to augment human memory. In a 1945 essay — As We May Think — he elaborated on this as transforming the information explosion into a knowledge explosion so as to make knowledge more accessible in helping fix societal problems. Doug Engelbart, the famous SRI computer pioneer, picked up on Bush’s ideas and made it a lifelong mission to augment collective human intelligence.
Now, almost a century after Bush’s nascent idea, developing “augmented knowledge systems” is finally realizable. A common knowledge base accessible everyone, anywhere, on-demand. More important, as both Bush and Engelbart understood, doing so is exceedingly desirable — socially, politically and economically. And, I would add, inevitable.
To be all you can be
Each of us are born as an incredible learning machine. Whatever our end goal might be, a central idea underlying our learning and education is to acquire knowledge — an explicit or tacit familiarity, awareness or understanding of facts, information, and or skills.
Ultimately, what and how much we actually learn and know impacts our opportunities and outcomes in life and, perhaps, societal development. So, the importance of a quality education and our diligence in its pursuit cannot be overstated. But, importantly, there are different routes to this end state.
For example, I think of myself as a reasonably knowledgeable person. However, like many people, most of my formal education — from primary school through undergraduate college work— was, at best, only marginally helpful in my development. Also, like many others who lacked a guiding parent or mentor, my default was to follow my own interests and curiosity.
Fortunately, early on I discovered libraries and, more importantly, librarians. With their help I became an autodidact — a self-taught acquirer of knowledge. While a challenging and risky approach, it has served me well.
Yet, I know, with certainty, that I do not now know as much as I might like and need. More importantly, that I will never be able to learn and know enough. None of us can, and that is a problem.
Know what I mean?
Knowing enough is important because, as we all know, “Knowledge is power.” Moreover, the corollary, the absence or lack of knowledge, implies impotence, weakness or vulnerability, as reflected in the aphorism, “What you do not know can hurt you.”
But, the unspoken reality is that, in today’s world, none of us really know much at all. So we are all actually far more vulnerable than we consciously realize.
For example, it is estimated that 90 percent of all the scientists that have ever lived are alive and working now. That means that both the sheer quantity of new knowledge is growing fast and some sizeable percentage of past knowledge is now wrong or needs updating.
The point is that the world is growing more complex. So, knowing what we need-to-know is growing ever more challenging and problematic.
For example, ignorance of the law is a legal principle that says a person may not escape liability for violating a law merely because he or she was unaware of its existence or content. Yet, every day thousands of new laws are written and thousands of court decisions alter the interpretation of past laws. Thus, legally speaking, we are all screwed!
Penny for your thoughts
What we are experiencing, albeit subtly, is a rapid growth in “knowledge asymmetry.” In other words, the gap between what we know and do not know is growing rapidly, which makes us increasingly vulnerable to unexpected problems.
Of course, complicating our knowledge problem is the fact that our biology, interests and available time all conspire to limit what we can know. For example, reading a book a week for life only amounts to 3,000 books out of some 143 million unique titles in the world. And, of course, the time and mind-space problem grows dramatically worse with the Internet.
The basic reality is that none of us can be knowledgeable in more than a couple of topics. Or, if you prefer, we are all actually exceedingly ignorant in almost every topic area. So, as for knowledge being power, generally we are all in trouble.
Less than two but more than one
Personally, not knowing things I should or would like to know is a daily irritant and frustration. This feeling is exacerbated whenever I learn something after it would have been useful to have known it.
But the absolute worst feeling is making an argument for or against something only to learn later it was erroneous because my own knowledge base was deficient. As an autodidact, whenever this occurs I mentally beat myself up. Why did I not accurately know what I should have been able to know?
Of course, in no way is this deficient knowledge base problem unique to me.
From the kitchen table, to the office, to the courtroom, to the barroom stool, to online chats all of us have experienced this problem — and more than once. Similarly, everyone from kids and their school work, to people working on business or professional projects, to families and their personal issues, concerns or crises have all found themselves in urgent need of accurate, reliable knowledge they lacked.
Then, there is the cruel irony of death. While much of what each of us learns is redundant, we nonetheless spend a lifetime building up our knowledge base only to lose it all. To repurpose an often used phrase, “a mind is a terrible thing to waste.”
The point is that the way we learn knowledge and the historical approach to building our knowledge base — both individually and collectively — is obsolete, deficient and often counterproductive.
As the worm turns
The problem of knowing what we need-to-know is exacerbated whenever there is a tacit or intentional reluctance to share knowledge.
This contrived knowledge asymmetry occurs most often when someone or organization believes that the withholding of some knowledge will afford them some advantage, edge, leverage, benefit or power. Of course, this is neither new nor uncommon.
The fact is that consciously creating knowledge asymmetries actually began in earnest with the introduction of writing. Whether for religious, economic, or political purposes, writing first enabled some individuals to inventory and record relationships, events and things — an external knowledge base — to the exclusion and or at the expense of others. So, it is not coincidental that this era gave rise to the phrase knowledge is power.
Consequently, when printing first blossomed in the 15th century the initial assumption was that it would enhance the power of the existing institutions. Instead, the widespread availability of knowledge enabled a personal and collective comparison of the existing rhetoric and reality in all topic areas.
The result was a general reduction in knowledge asymmetries. Said differently, easy and widespread access to a common knowledge base better leveled civilization’s intellectual playing field in all topic areas.
Two steps forward one step back
Nonetheless, over the years, various individuals, groups and organizations — e.g., education, trades, business, finance, legal, military, and governments — regularly sought to create new ways to foster knowledge asymmetries and advance their respective self-interest advantages.
To counter these new asymmetry efforts other groups and organizations sought to formulate new methods and actions aimed at reducing their impact: e.g., universal public education, growth in public libraries, court actions against segregation and discrimination, consumer product and service reviews and ratings, creation of governmental agencies like the federal trade commission, food and drug administration, consumer financial protection bureau, the freedom of information act, and so on.
Obviously, some knowledge asymmetries will always exist. However, it also likely that failing to significantly reduce knowledge asymmetries wherever possible only serves to increase counterproductive social and political polarization and conflict.
The point is, setting aside subjective and philosophical issues, there should be more, better and simpler ways to regularly obtain reliable knowledge on-demand. Fortunately, we have already glimpsed it.
The future is here, just not evenly distributed
Just as printing subtly reduced knowledge asymmetries, the Internet is in the process of subtly doing the same thing. And, like printing, the reduction in knowledge asymmetries from the Internet will eventually be globally viral, ubiquitous and permanently change in the course of civilization.
The key to the realization of this capability will be new filters to efficiently navigate our growing abundance of available online content, and various guides to using reliable knowledge more efficaciously. That is, augmented knowledge systems akin to those envisioned Bush and Engelbart.
The camel’s nose under the Internet tent and pointing to where we are going was the proliferation of price comparison tools. In that instance, what started as a simple online version of printed price comparison lists quickly evolved into dynamic algorithmic tools highlighting the cheapest price available in a given product or service category — in real-time and on-demand.
As these services expanded to incorporate more vertical categories of products and services they ultimately facilitated “showrooming.”
Showrooming is a simple idea with world changing consequences. Essentially, it enables any user to instantly compare prices inside physical stores or anywhere else with competing prices anywhere online and offline.
Said differently, the inherent attractiveness of showrooming is how the common knowledge base reduces knowledge asymmetries — in real-time and on-demand.
The result has been to drive down prices on virtually every product and service touched. Moreover, it is permanently changing the nature of the previously asymmetric relationship between business and consumers. But this is just the beginning.
New augmented knowledge systems
At present, there are two obstacles to reducing asymmetries for knowledge content on the Internet in real-time and on-demand:
- databases of reliable knowledge
- ubiquitous ambient knowledge access
- Databases of reliable knowledge— that separate the most accurate and reliable knowledge content from the rest to augment what we need but do not know.
Like anything else in life, there are content providers and knowledge content that is superior to the rest. The central characteristic of price comparison services for showrooming is a database of the best prices for each product and service that is regularly updated. But there is no analogous database for knowledge content on the Internet.
The basic problem is that today’s Internet search engines select content from the global database of information based primarily on popularity and longevity — not the quality, reliability or accuracy of the content. Thus, the onus is on each of us to know what content is reliable and accurate.
But, absent sufficient time, effort, skill and know-how to identify a superior content for any and every topic needed, knowledge asymmetry continues. Worse, there is now an epidemic of bad and erroneous information, misinformation and disinformation infecting civilization and effectively making us less knowledgeable and more polarized.
So, the first key is to reducing knowledge asymmetries is the development of dynamic, superior knowledge content databases.
To this end, some believe machine learning leading to AI systems can do this. But, as everyone in tech industry acknowledges, albeit reluctantly, for the foreseeable future there are two problems preventing this from occurring.
First, comparing numbers reflecting prices is easy for computer algorithms. However, there is no known algorithm capable of providing the needed absolute and unambiguous understanding of the nuances in written and or spoken language.
In other words, building a highly versatile machine learning solution for understanding the diversity of users and their respective spoken and or written intent is exceedingly difficult. The simple fact is that the same text often has very different correct meanings. As Google itself notes:
One of the main problems that makes <text> parsing so challenging is that human languages show remarkable levels of ambiguity. It is not uncommon for moderate length sentences — say 20 or 30 words in length — to have hundreds, thousands, or even tens of thousands of possible syntactic structures. A natural language parser must somehow search through all of these alternatives, and find the most plausible structure given the context — Google Research Blog (5–12–16)
Second, when such a machine learning program finally does exist, the criteria used to build a knowledge database will be different for each different topic area. Consequently, distinct, separate criteria and specialized training will be needed for each distinct topic database.
Thus, for the foreseeable future, to build and dynamically maximize the quality of a knowledge database will require human curation. Moreover, it will require a combination of expert and public inputs with strict adherence to key, relevant topic related criteria and absolute quality control systems.
That said, accessing knowledge content via traditional text and keyboards mainly requires a discrete, trustworthy service. This is straightforward value-added system that directly compares listed search results against a curated knowledge database to instantly identify superior content within normal search results without altering the user’s behavior in any way. (Full disclosure, this is what my company provides.)
The great irony as a friend and mentor, Keith Teare, has pointed out, is that venture capitalist (or as the engagingly smart, Curtis Feeny of Voyager Capital describes them, chicken-capitalists) have a myopic view of what is valuable. So, there is a reluctance to support database development, regardless of its ultimate absolute value. But, in the context of the second obstacle needed to reduce knowledge asymmetries, this will soon change.
Emergence of ambient augmented knowledge systems
- Ubiquitous ambient knowledge access — is critical because the degree of effort required to obtain reliable knowledge determines the usefulness and value of ambient systems.
What makes the current investor orientation all the more ironic is that third-party databases are what feed voice activated virtual assistants (e.g., Siri, Echo, Viv). More to the point, voice activated virtual assistants are probably the biggest money maker on the tech horizon. Yet, these systems can never go mainstream and realize their economic potential without reliable knowledge content databases.
Said differently, no one wants to be talking to a machine that is telling you unreliable or outright wrong knowledge content, especially if it will affect something important in your life.
In other words, the essential component for any mainstream success of virtual assistants is a need-to-know knowledge database. A database with content people know they can trust and be confident is always accurate and reliable.
Once there is any mainstream acceptance of virtual assistants they will quickly go viral and become ubiquitous, yet invisible, in every environment and setting. At that point they become ambient augmenting knowledge systems. This will result in a dramatic reduction of knowledge asymmetries on a societal scale. Thus, in effect, we will all become autodidacts.
So, soon, the early, prescient visions of Vannevar Bush and Doug Engelbart will finally be realized. More importantly, civilization will forever be the real beneficiary.
If you enjoyed this post, and want to share it, please hit “Recommend” below. Thanks! It helps spread these ideas!
In any case, may you live long and prosper.