Saturday, February 11, 2012
In recent years I have been working on two major concepts: first, the connectivist theory of online learning, which views learning as a network process; and second, the massive open online course, or MOOC, which is an instantiation of that process. These, however, represent only the most recent of what can be seen as a series of 'generations' of e-learning. In this talk I describe these generations and discuss how they led to, and are a part of, the most recent work in online learning.
Thank you for welcoming me to your conference.
The theme I would like to explore today concerns the growth and development of our idea of online learning, or as it is sometimes called, e-learning. What I would like to do is to describe a series of 'generations' of technologies and approaches that have characterized the development of online learning over the years. These generations of have informed the shape of online learning as it exists today, and will help us understand something of the direction it will take in the future.
These generations span more than a 20-year period. Indeed, there may even be described a 'generation zero' that predates even my own involvement in online learning. This generation is characterized by systems such as Plato, and represents the very idea of placing learning content online. This includes not only text but also images, audio, video and animations. It also represents, to a degree, the idea of programmed learning. This is the idea that computers can present us with content and activities in a sequence determined by our choices and by the results of online interactions, such as tests and quizzes. We have never wandered far from this foundational idea, not even in the 21st century. And it continues to be the point of departure for all subsequent developments in the field of online learning.
For me, 'generation one' consists of the idea of the network itself. My first work in the field of online learning was to set up a bulletin board system, called Athabaska BBS, in order to allow students from across the province to communicate with me online. It was also the time I first began using email, the time I began using the Usenet bulletin Board system, and the time I first began using online information systems such as Gopher. The process of connecting was involved and complex, requiring the use of modems and special software.
As generation one developed, generation zero matured. The personal computer became a tool anyone could use to create and store their own content. Commercial software came into existence, including both operating systems and application programs such as spreadsheets, word processors, and database tools. Content could be created in novel ways - the 'mail merge' program, for example, would allow you to print the same letter multiple times, but each with a different name and address drawn from a database.
The next generation takes place in the early 1990s and is essentially the application of computer games to online learning. These games were in the first instance text-based and very simple. But they brought with them some radical changes to the idea of learning itself.
One key development was the idea that multiple people could occupy the same online 'space' and communicate and interact with each other. This development coincided with the creation of IRC - inter-relay chat - and meant that you were in real time communication with multiple people around the world. But more: the gaming environment meant you could do things with other people - explore terrain, solve puzzles, even fight with them.
Another key idea was the design of the gaming space itself. Early computer games (and many early arcade games) were designed like programmed learning: they were like a flow chart, guiding you through a series of choices to a predetermined conclusion. But the online games were much more open-ended. Players interacted with the environment, but the outcome was not predetermined. At first it was created by chance, as in the rolling of dice in a Dungeons and Dragons game. But eventually every game state was unique, and it was no longer possible to memorize the correct sequence of steps to a successful outcome.
The third element was the technology developed to enable that which we today call object oriented programming. This changed the nature of a computer program from a single entity that processed data to a collection of independent entities - objects - that interacted with each other: they could send messages to each other to prompt responses, one could be 'contained' in another, or one could be 'part' of another. So a game player would be an object, a monster would be an object, they would be contained in a 'room' that was also an object, and gameplay consisted of the interactions of these objects with each other in an unplanned open-ended way.
During the development of this second generation we saw the consolidation of computer-based software and content, and the commercialization of the network itself. The many brands we saw in the 80s - Atari, Amiga, Tandy, IBM, and many more - coalesced into the now familiar Mac-PC divide. A few major software developers emerged, companies like Microsoft and Corel. Computers became mainstream, and became important business (and learning) tools.
Meanwhile, the world of networks began to commercialize. Commercial bulletin board services emerged, such as Prodigy, AOL, GEnie and Compuserv. And the first local internet service providers came into being. Networking became the way important people connected, and communities like the WELL began to define a new generation of thought leaders.
You can begin to see a pattern developing here. Through the first three generations, a familiar process of innovation occurs: first the development and piloting of the technology (which is also when the open source community springs up around it), then the commercialization of the technology, then the consolidation of that commercial market as large players eliminate weaker competitors.
The next generation sees the development of the content management system, and in learning, the learning management system.
Both of these are applications developed in order to apply the functionality developed in generation zero - content production and management - to the platform developed in generation one - the world wide web. The first content management systems were exactly like mail merge, except instead of printing out the content, they delivered it to the remote user (inside a computer program, the commands are exactly the same - 'print' is used to print data to a page, print data to a file, or print data to the network).
Early learning management systems were very easy to define. They consisted of a set of documents which could be merged with a list of registered users for delivery. They also supported some of the major functions of networks: bulletin boards, where these users could post messages to each other, chat rooms, where they could occupy the same online space together, and online quizzes and activities, where they could interact with the documents and other resources.
It is interesting to me to reflect that the major debates about online learning around this time centered on whether online learning would be mostly about online content - that is, reflective of generation zero - or mostly about online interaction - that is, reflective of generation one. I remember some teachers in Manitoba swearing by the interaction model, and using a bulletin-board style application called FirstClass - eschewing to more content-based approach I was favouring at the time.
Learning management systems drew a great deal from distance learning. Indeed, online was (and is still) seen as nothing more than a special type of distance learning. As such, they favoured a content-based approached, with interaction following secondarily. And a very standard model emerged: present objectives, present content, discuss, test. More advanced systems attempted to replicate the programmed learning paradigm. The Holy Grail of the day was adaptive learning - a system which would test you (or pretest you) to determine your skill level, then deliver content and activities appropriate to that level.
Despite its now-apparent shortcomings, the learning management system brought some important developments to the field.
First, they brought the idea that learning content could be modularized, or 'chunked'. This enabled a more fine-grained presentation of learning content than traditional sources such as textbooks and university courses. Shorter-form learning content is almost ubiquitous today.
Second, it created the idea that these content modules or chunks were sharable. The idea that books or courses could be broken down into smaller chunks suggested to people that these chunks could be created in one context and reused in another context.
And third, they brought together the idea of communication and content in the same online environment. The learning management system became a place where these smaller content objects could be presented, and then discussed by groups of people either in a discussion board or in a live chat.
These were the core elements of learning management technology, and a generation of online learning research and development centered around how content should be created, managed and discussed in online learning environments. People discussed whether this form of learning could be equal to classroom learning, they discussed the methodology for producing these chunks, and they discussed the nature, role and importance of inline interaction.
Around this time as well an ambitious program began in an effort to apply some of the generation two principles to learning management systems (and to content management in general). We came to know this effort under the heading of 'learning objects'. In Canada we had something called the East-West project, which was an attempt to standardize learning resources. The United States developed IMS, and eventually SCORM. Most of the work focused on the development of metadata, to support discoverability and sharing, but the core of the program was an attempt to introduce second generation technology - interactive objects - to learning and content management.
But it didn't take hold. To this day, the learning management system is designed essentially to present content and support discussion and activities around that content. We can understand why when we look at the development of the previous generations of online learning.
By the time learning management systems were developed, operating systems and application programs, along with the content they supported, were enterprise software. Corporations and institutions supported massive centralized distributions. An entire college or university would standardize on, say, Windows 3.1 (and very few on anything else). 'Content' became synonymous with 'documents' and these documents - not something fuzzy like 'objects' - were what would be created and published and shared.
The network was by this time well into the process of becoming consolidated. Completely gone was the system of individual bulletin board services; everything now belonged to one giant network. Telecoms and large service providers such as AOL were coming to dominate access. The internet standardized around a document presentation format - HTML - and was defined in terms of websites and pages, constituting essentially a simplified version of the content produced by enterprise software. The same vendors that sold these tools - companies like Microsoft and Adobe - sold web production and viewing tools.
Probably the most interesting developments of all at the time were happening outside the LMS environment entirely. The tools used to support online gaming were by this time becoming commercialized. It is worth mentioning a few of these. New forms of games were being developed and entire genres - strategy games, for example, sports games, and first-person shooters - became widely popular.
Though gaming remained a largely offline activity, online environments were also beginning to develop. One of the first 3D multi-user environments, for example, was Alpha Worlds. This was followed by Second Life, which for a while was widely popular. Online gaming communities also became popular, such as the chess, backgammon and card playing sites set up by Yahoo. And of course I would be remiss if I didn't mention online gambling sites.
As I mentioned, these developments took place outside the LMS market. The best efforts of developers to incorporate aspects of gaming - from object oriented learning design to simulations and gaming environments to multi-user interactions - were of limited utility in learning management systems. LMSs were firmly entrenched in the world of content production, and to a lesser extent the world of networked communication.
This leads us next to the fourth generation, paradoxically called web 2.0 - and in the field of online learning, e-learning 2.0.
The core ideas of web 2.0 almost defy description in previous terminology. But two major phenomena describe web 2.0 - first, the rise of social networks, and second, the creation of content and services that can interact with those networks. Web 2.0 is sometimes described as the 'web as a platform' but it is probably more accurate to see it as networking being applied to data (or perhaps data being applied to networking).
The core technology of web 2.0 is social software. We are most familiar with social software through brand names like Friendster, MySpace, Twitter, Linked In, Facebook, and most recently, Google+. But if we think for a moment about what social software is, it is essentially the migration of some of your personal data - like your mailing list - to a content management system on the web. These systems then leverage that data to create networks. So you can now do things online - like send the same message to many friends - that you could previously only do with specialized applications.
E-learning 2.0 is the same idea applied to e-learning content. I am widely regarded as one of the developers of e-learning 2.0, but this is only because I recognized that a major objective of such technologies as learning objects and SCORM was to treat learning resources as data. The idea was that each individual would have available online the same sort of content authoring and distribution capabilities previously available only to major publishers. And these would be provided online.
E-learning 2.0 brings several important developments to the table.
First, it brings in the idea of the social graph, which is essentially the list of people you send content to, and the list of people who send you content, and everyone else's list, all in one big table. The social graph defines a massive communications network in which people, rather than computers, are the interconnected nodes.
Second, it brings in the idea of personal publishing. The beginning of web 2.0 is arguably the development of blogging software, which allowed people to easily create web content for the first time. But it's also Twitter, which made creating microcontent even easier, and YouTube, which allowed people to publish videos, and MySpace, which did the same for music, and Facebook and Flickr, which did the same for photos.
Third, it brings in the idea of interoperability, first in the form of syndication formats such as RSS, which allow us to share our content easily with each other, but also later in the form or application programming interfaces, which allow one computer program on one website to communicate with another program on another website. These allow you to use one application - your social network platform, for example - to use another application - play a game, edit content, or talk to each other.
And fourth, it brings us the idea of platform-independence. Web 2.0 is as much about mobile computing as it is about social software. It is as much about using your telephone to post status updates or upload photos as it is about putting your phonebook on a website. Maybe even more so.
What made web 2.0 possible? In a certain sense, it was the maturation of generation 0, web content and applications. After being developed, commercialized and consolidated, these became enterprise services. But as enterprises became global, these two become global, and emerged out of the enterprise to become cloud and mobile contents and applications.
Some of the major social networking sites are actually cloud storage sites - YouTube and Flickr are the most obvious examples. Some are less obvious, but become so when you think about it - Wikipedia, for example. Other cloud storage sites operate behind the scenes, like Internet Archive and Amazon Web Services. And there are cloud services, like Akamai, that never reach the mainstream perception.
These cloud services developed as a result of enterprise networking. On the research side, high-speed backbones such as Internet 2 in the U.S. and CA*Net 3 in Canada virtually eliminated network lag even for large data files, audio and video. Similar capacities were being developed for lease by the commercial sector. And the now-consolidated consumer market now began to support always-on broadband capacity through ASDL or cable internet services.
The consolidation of core gaming technologies took place largely behind the scenes. This era sees the ascendance of object-oriented coding languages such as Java and dot Net. The open-ended online environment led to massive multiplayer online games such as Eve and World of Warcraft. In learning we see the emergence of major simulation developers such as CAE and conferencing systems such as Connect, Elluminate, and Cisco. These have become dominant in the delivery of online seminars and classes.
Content management services, meanwhile, were increasingly commercialized. We saw the emergence of Blackboard and WebCT, and on the commercial side products like Saba and Docent. Google purchased Blogger, Yahoo purchased Flickr, and even the world of open source systems came to be dominated by quasi-commercial enterprises. Innovators moved on and began to try radical new technologies like RSS and AJAX, Twitter and Technorati. Today we think of social networking in terms of the giants, but when it started in the mid-2000s the technology was uncertain and evolving. In education, probably the major player from this era was Elgg, at that time and still to this day a novel technology.
Today, of course, social networking is ubiquitous. The major technologies have been commercialized and are moving rapidly toward commodification and enterprise adoption. The ubiquity of social networking came about as a result of the commercialization of content management services. A new business model has emerged in which providers sell information about their users to marketing agencies. The proliferation of social networking sites has now been reduced to a few major competitors, notably YouTube, Facebook and Twitter. The providers of search and document management services - Yahoo, Microsoft, Apple and Google - have their own social networks, but these are also-rans. Hence when people speak of 'social network learning' they often mean 'using Facebook to support learning' or some such thing.
This is the beginning of the sixth generation, a generation characterized by commercialized web 2.0 services, a consolidation of the CMS/LMS market, the development of enterprise conferencing and simulation technology, cloud networking and - at last - open content and open operating systems.
Now before the Linux advocates lynch me, let me say that, yes, there have always been open operating systems. But - frankly - until recently they have always been the domain of innovators, enthusiasts and hobbyists. Not mainstream - not, say, running underlying major commercial brands, the way Linux now underlies Apple's OSX, and not widely used, say, the way Android powers a large percentage of mobile phones.
So that's the history of online learning through five generations, but it is also a listing of the major technologies that form the foundation for sixth-generation e-learning, which I would characterized by the Massive Open Online Course.
Let me spend a few moments talking about the development of the MOOC model.
When George Siemens and I created the first MOOC in 2008 we were not setting out to create a MOOC. So the form was not something we designed and implemented, at least, not explicitly so. But we had very clear ideas of where we wanted to go, and I would argue that it was those clear ideas that led to the definition of the MOOC as it exists today.
There were two major influences. One was the beginning of open online courses. We had both seen them in operation in the past, and had most recently been influenced by Alec Couros's online graduate course and David Wiley's wiki-based course. What made these courses important was that they invoked the idea of including outsiders into university courses in some way. The course was no longer bounded by the institution.
The other major influence was the emergence of massive online conferences. George had run a major conference on Connectivism, in which I was a participant. This was just the latest in a series of such conferences. Again, what made the format work was that the conference was open. And it was the success of the conference that made it worth considering a longer and more involved enterprise.
We set up Connectivism and Connective Knowledge 2008 (CCK08) as a credit course in Manitoba's Certificate in Adult Education (CAE), offered by the University of Manitoba. It was a bit of Old Home Week for me, as Manitoba's first-ever online course was also offered through the CAE program, Introduction to Instruction, designed by Conrad Albertson and myself, and offered by Shirley Chapman.
What made CCK08 different was that we both decided at the outset that it would be designed along explicitly connectivist lines, whatever those were. Which was great in theory, but then we began almost immediately to accommodate the demands of a formal course offered by a traditional institution. The course would have a start date and an end date, and a series of dates in between, which would constitute a course schedule. Students would be able to sign up for credit, but if they did, they would have assignments that would be marked (by George; I had no interest in marking).
But beyond that, the course was non-traditional. Because when you make a claim like the central claim of connectivism, that the knowledge is found in the connections between people with each other and that learning is the development and traversal of those connections, then you can't just offer a body of content in an LMS and call it a course. Had we simply presented the 'theory of connectivism' as a body of content to be learned by participants, we would have undercut the central thesis of connectivism.
This seems to entail offering a course without content - how do you offer a course without content? The answer is that the course is not without content, but rather, that the content does not define the course. That there is no core of content that everyone must learn does not entail that there is zero content. Quite the opposite. It entails that there is a surplus of content. When you don't select a certain set of canonical contents, everything becomes potential content, and as we saw in practice, we ended up with a lot of content.
Running the course over fourteen weeks, with each week devoted to a different topic, actually helped us out. Rather than constrain us, it allowed us to mitigate to some degree the effects an undifferentiated torrent of content would produce. It allowed us to say to ourselves that we'll look at 'this' first and 'that' later. It was a minimal structure, but one that seemed to be a minimal requirement for any sort of coherence at all.
Even so, as it was, participants complained that there was too much information. This led to the articulation of exactly what connectivism meant in a networked information environment, and resulted in the definition of a key feature of MOOCs. Learning in a MOOC, we advised, is in the first instance a matter of learning how to select content.
By navigating the content environment, and selecting content that is relevant to your own personal preferences and context, you are creating an individual view or perspective. So you are first creating connections between contents with each other and with your own background and experience. And working with content in a connectivist course does not involve learning or remembering the content. Rather, it is to engage in a process of creation and sharing. Each person in the course, speaking from his or her unique perspective, participates in a conversation that brings these perspectives together.
Why not learn content? Why not assemble a body of information that people would know in common? The particular circumstances of CCK08 make the answer clear, but we can also see how it generalizes. In the case of CCK08, there is no core body of knowledge. Connectivism is a theory in development (many argued that it isn't even a theory), and the development of connective knowledge even more so. We were hesitant to teach people something definitive when even we did not know what that would be.
Even more importantly, identifying and highlighting some core principles of connectivism would undermine what it was we thought connectivism was. It's not a simple set of principles or equations you apply mechanically to obtain a result. Sure, there are primitive elements - the component of a connection, for example - but you move very quickly into a realm where any articulation of the theory, any abstraction of the principles, distorts it. The fuzzy reality is what we want to teach, but you can't teach that merely by assembling content and having people remember it.
So in order to teach connectivism, we found it necessary for people to immerse themselves in a connectivist teaching environment. The content itself could have been anything - we have since run courses in critical literacies, learning analytics, and personal learning environments. The content is the material that we work with, that forms the creative clay we use to communicate with each other as we develop the actual learning, the finely grained and nuanced understanding of learning in a network environment that develops as a result of our working within a networked environment.
In order to support this aspect of the learning, we decided to make the course as much of a network as possible, and therefore, as little like an ordered, structured and centralized presentation as possible. Drawing on work we'd done previously, we set up a system whereby people would use their own environments, whatever they were, and make connections between each other (and each other's content) in these environments.
To do this, we encouraged each person to create his or her own online presence; these would be their nodes in the course networks. We collected RSS feeds from these and aggregated them into a single thread, which became the course newsletter. We emphasized further that this thread was only one of any number of possible ways of looking at the course contents, and we encouraged participants to connect in any other way they deemed appropriate.
This part of the course was a significant success. Of the 2200 people who signed up for CCK08, 170 of them created their own blogs, the feeds of which were aggregated with a tool I created, called gRSShopper, and the contents delivered by email to a total of 1870 subscribers (this number remained constant for the duration of the course). Students also participated in a Moodle discussion forum, in a Google Groups forum, in three separate Second Life communities, and in other ways we didn't know about.
The idea was that in addition to gaining experience making connections between people and ideas, participants were making connections between different systems and places. What we wanted people to experience was that connectivism functions not as a cognitive theory - not as a theory about how ideas are created and transmitted - but as a theory describing how we live and grow together. We learn, in connectivism, not by acquiring knowledge as though it were so many bricks or puzzle pieces, but by becoming the sort of person we want to be.
In this, in the offering of a course such as CCK08, and in the offering of various courses after, and in the experience of other people offering courses as varied as MobiMOOC and ds106 and eduMOOC, we see directly the growth of individuals into the theory (which they take and mold in their own way) as well as the growth of the community of connected technologies, individuals and ideas. And it is in what we learn in this way that the challenge to more traditional theories becomes evident.
Now I mentioned previously that the MOOC represents a new generation of e-learning. To understand what that means we need to understand what the MOOC is drawing from the previous generations, and what the MOOC brings that is new.
Let me review:
Generation 0 brings us the idea of documents and other learning content, created and managed using application programs. In this the sixth generation of such technologies we have finally emerged into the world of widespread free and open online documents and application programs. The ability to read and write educational content, to record audio and make video, is now open to everybody, and we leverage this in the MOOC. But this is not what makes the MOOC new.
Additionally, a fundamental underlying feature of a connectivist course is the network, which by now is in the process of becoming a cloud service. WiFi is not quite ubiquitous, mobile telephony is not quite broadband, but we are close enough to both that we are connected to each other on an ongoing basis. The MOOC leverages the network, and increasingly depends on ubiquitous access, but this is not what makes the MOOC new.
The MOOC as we have designed it also makes use of enterprise 'game' technology, most specifically the conferencing system. Elluminate has been a staple in our courses. We have also used - and may well use again in the future - environments such as Second Life. Some other courses, such as the Stanford AI course, have leveraged simulations and interactive systems. Others, like ds106, emphasize multimedia. Using these and other immersive technologies, the MOOC will become more and more like a personal learning environment, but this is not what makes the MOOC unique.
The MOOC also makes explicit use of content management systems. The early MOOCs used Moodle; today we encourage participants to use personal content management systems such as WordPress and Blogger. The gRSShopper environment itself is to a large degree a content management system, managing a large store of user contributions and facilitator resources. But clearly, the element of content management is not what makes the MOOC new.
And the MOOC makes a lot of use of commercial social networking services. Twitter feeds and the Facebook group are major elements of the course. Many students use microblogging services like Posterous and Tumblr. Like membership in a social network, membership in the course constitutes participation in a large graph; contents from this graph are aggregated and redistributed using social networking channels and syndication technologies. But many courses make use of social networks. So that is not what makes a MOOC unique.
So what's new? I would like to suggest that the MOOC adds two major elements to the mix, and that it is these elements that bear the most investigation and exploration.
First, the MOOC brings the idea of distributed technology to the mix. In its simplest expression, we could say that activities do not take place in one central location, but rather, are distributed across a large network of individual sites and services. The MOOC is not 'located' at cck12.mooc.ca (or at least, it's not intended to me) - that is just one nexus of connected sites.
In fact, it is the idea of distributed knowledge that is introduced by the MOOC again, and the means of learning is really involved with this idea. When you learn as a network, you cannot teach one fact after another. Each fact is implicated with the others. You cannot see a single fact, even if you extract a fact from the data, because it would be only one abstraction, an idealization, and not more true that the identification of regularities in the data - and learning becomes more like a process to create landforms, and less like an exercise of memory. It is the process of pattern recognition that we want to develop, and not the remembering of facts.
Accordingly, the second element the MOOC brings to the mix revolves around the theory of effective networks. More deeply, the MOOC represents the instantiation of four major principles of effective distributed systems. These principles are, briefly, autonomy, diversity, openness and interactivity.
For example, it is based on these principles that we say that it is better to obtain many points of view than one. It is based on these principles that we say that the knowledge of a collection of people is greater than just the sum of each person’s knowledge. It is based on these principles that we argue for the free exchange of knowledge and ideas, for open education, for self-determination and personal empowerment.
These four principles form the essence of the design of the network - the reason, for example, we encourage participants to use their preferred technology (it would be a lot easier if everybody used WordPress).
We are just now as a community beginning to understand what it means to say this. Consider 'learning analytics', for example, which is an attempt to learn about the learning process by examining a large body of data.
What is learned in the process of learning analytics is not what is contained in individual bits of data - that would be ridiculous - but overall trends or patterns. What is learned, in other words, emerges from the data. The things we are learning today are very simple. In the future we expect to learn things that are rather more subtle and enlightening.
Let me now say a few words in closing about Generation 6 and beyond.
From my perspective, the first three generations of e-learning (and the web generally) represent a focus on documents, while the second three represent a focus on data. Sometimes people speak of the second set as a focus on the Semantic Web, and they would not be wrong. Data does not stand alone, the way documents do; the representation of any object is connected to the representation of any number of other objects, through shared features or properties, or by being related by some action or third party agency.
Indeed, if the first three generations are contents, networks and objects respectively, the second three generations are those very same things thought of as data: the CMS is content thought of as data, web 2.0 is the network thought of as data, and the MOOC is the environment thought of as data. So what comes after data is pretty important, but I would say, it is also to a certain degree knowable, because it will have something to do with content, the network, and the environment.
Here's what I think it will be - indeed, here's what I've always thought it would be. The next three generations of web and learning technology will be based on the idea of flow.
Flow is what happens when your content and your data becomes unmanageable. Flow is what happens when all you can do is watch it as it goes by - it is too massive to store, it is too detailed to comprehend. Flow is when we cease to think of things like contents and communications and even people and environments as things and start thinking of them as (for lack of a better word) media - like the water in a river, like the electricity in our pipes, like the air in the sky.
The first of these things that flow will be the outputs of learning (and other) analytics; they will be the distillation of the massive amounts of data, presented to us from various viewpoints and perspectives, always changing, always adapting, always fluid.
Inside the gRSShopper system I am working toward the development of the first sort of engines that capture and display this flow. gRSShopper creates a graph of all links, all interactions, all communications. I don't know what to do with it yet, but I think that the idea of comprehending the interactions between these distributed systems in a learning network is an important first step to understanding what is learned, how it is learned, and why it is learned. And with that, perhaps, we can take our understanding of online learning a step further.
But that, perhaps, may take the efforts of another generation.