Elearning is a complex subject that ranges from theories of learning to instructional design to human-computer interfaces. This memo outlines the main issues and concerns with quotes and citations from various websites.

WHAT'S AN INSTRUCTIONAL SYSTEM:
An instructional system is an arrangement of resources and procedures to promote learning. Instructional design is the systematic process of developing instructional systems and instructional development is the process of implementing the system or plan.
http://www.umich.edu/~ed626/define.html

Instructional technology is a complex, integrated process involving people, procedures, ideas, devices, and organization, for analyzing problems, and devising, implementing, evaluating and managing solutions to those problems, in situations in which learning is purposive and controlled.
http://courses.educ.ksu.edu/EDETC886/ProSemB/
Chandima/AECTdefinition.html

A BIT OF EDUCATION THEORY:
Then, at the turn of the 20th century John Dewey presented several tenets of the philosophy of education which promoted the idea that learning occurs best when married with doing, rather than rote regurgitation of facts.

Ralph Tyler's work a decade before WWII indicated that objectives were most useful to instructional developers if written in terms of desired learner behaviors.

... programmed instruction emphasizes the formulation of behavioral objectives, breaking instructional content into small units and rewarding correct responses early and often.

The advent of new media, such as the Internet and hypermedia, has brought about not only technological innovations, but also coupled these with new ways of approaching learning and instruction. As opposed to the behavioralist perspective that emphasizes learning objectives, the constructivist approach holds that learners construct their understanding of reality from interpretations of their experiences.
 http://www.pignc-ispi.com/articles/education/
brief%20history.htm

… initial knowledge acquisition is perhaps best served by classical instruction with predetermined learning outcomes, sequenced instructional interaction and criterion-referenced evaluation while the more advanced second phase of knowledge acquisition is more suited to a constructivist environment.

Jonnassen ... identified the following types of learning and matched them with what he believes to be appropriate learning theory approaches.
1. Introductory Learning - learners have very little directly transferable prior knowledge about a skill or content area. They are at the initial stages of schema assembly and integration. At this stage classical instructional design is most suitable because it is predetermined, constrained, sequential and criterion-referenced. The learner can develop some anchors for further exploration.
2. Advanced Knowledge Acquisition - follows introductory knowledge and precedes expert knowledge. At this point constructivist approaches may be introduced.
3. Expertise is the final stage of knowledge acquisition - In this stage the learner is able to make intelligent decisions within the learning environment. A constructivist approach would work well in this case.
Having pointed out the different levels of learning, Jonassen stresses that it is still important to consider the context before recommending any specific methodology.

http://www.usask.ca/education/coursework/
802papers/mergel/brenda.htm

ROLE OF TECHNOLOGY:
There were debates over the role of technology: as tool or mechanism to promote solid instructional practices; as supplementary to the classroom: or as a replacement for the teacher. And of course, the never-ending question of acceptance by the educational establishment, the de-humanizing of education, …
http://www.quasar.ualberta.ca/EDIT572/572h2.html

One of the most useful tools for the constructivist designer is hypertext and hypermedia because it allows for a branched design rather than a linear format of instruction. Hyperlinks allow for learner control which is crucial to constructivist learning; however, there is some concerns over the novice learner becoming "lost" in a sea of hypermedia.
http://www.usask.ca/education/coursework/
802papers/mergel/brenda.htm

WHAT IT'S GOOD FOR:
Most of the survey participants agreed that e-learning is more effective for teaching the following topics: IT, basic business knowledge, new employee orientation, job specific skills, professional certification, and safety.

Classroom has proven to be the better tool for skill-based training and highly complex topics, where instructors need to see the puzzled looks of learners, and learners need to have discussions with peers. Topics identified as more suitable for classroom rather than for e-learning delivery include team building, professional effectiveness (time management, planning, stress management, active listening), mentoring, and leadership.
http://www.learningcircuits.org/2003/jan2003/nelson.htm

DESIGN PROCESS & CONCERNS:
Ideally, courseware is developed by teams including four types of people: an Instructional Designer, a subject specialist, a resource (graphic) designer and a programmer.

Phase I Design
State instructional goal
Perform Instructional analysis
Develop performance objectives
Develop testing strategies
Design instructional strategies
Phase II Pre-programming development
Develop flow charts and storyboards
Develop support materials
Design team review and revision
Phase III Development/Evaluation
Program first-draft materials
Perform formative evaluation (Within each phase there are feedbacks for revision)
-- Roblyer 1988
http://www.keele.ac.uk/depts/cs/Stephen_Bostock/
docs/atidass.htm

Singh described the ability to develop learning objects that consist of an individual unit of instruction (e.g. how to brew a great cup of coffee) which can be stored in a repository. They can then be called upon and assembled into larger "learning paths" (e.g. how to make a great breakfast) or served up individually to meet a specific information request. The learning object paradigm opens the door to true reusability of content, since content is no longer locked in monolithic course structures, Singh said.

… in the coming wave of learning objects-based eLearning: how does one preserve and leverage context while moving toward a modular approach to creating and delivering learning.
http://www.elearningforum.com/meetings/
2001/july/wdc1.htm

... in experiential instruction the computer simulates the subject matter, not the tutor, so that the learner interacts with the subject. In parallel, courseware can also provide a tutor role such as a coach, guide or expert demonstrating skills. … These experiential, performance-oriented approaches would work best where cognitive skills (rules) are being learned. Where declarative knowledge (concepts) are being learned there is still an issue as to how to stimulate appropriate cognitive processing and thus achieve deep (meaningful) rather than shallow (rote) learning.

Most courseware is interactive but the type of interaction should facilitate meaningful knowledge acquisition rather than rote or associative learning. One design strategy to acheive this is the 'electronic notebook' on which various forms of notetaking can be performed in response to interaction with the courseware. Learning activities need not take place on the screen but on paper, for example, or alternatively the medium for these learning strategies could be closely integrated with the courseware. … These broad principles must be augmented with more specific rules to generate designs of small sections of the courseware dealing with individual concepts or skills.
http://www.keele.ac.uk/depts/cs/Stephen_Bostock/
docs/atidass.htm

THE OTHER SIDE:
However, the "move to the Web" came with considerable downsides:
· Limited engineering technologies for Web server-based courseware shot down much of the interactive and instructional richness of Web courses as compared to the CD-ROM courses they were replacing.
· Bandwidth continued to be a limitation and prohibited media richness by shooting down much of the use of digital media. Even today, server latency and network congestion present barriers to seamless inclusion of streaming video in eLearning despite healthy gains in bandwidth.
· Ponderous wait times drastically interrupted and denigrated the learning experience.
The end result was a proliferation of dumbed-down courseware, even a preponderance of "page-turners" that learners tended to abandon before completion. Customers realized that eLearning held a lot of promise but not until something along the lines of the interactive, instructional, and media richness of CD-ROMs could be duplicated, albeit reinvented in the world of Web standards.

Pensare invented the "community" learning portal, which de-emphasized pre-built learning content in favor of collaborative experiences.
http://www.internettime.com/Learning/articles/DNA.pdf

GENERAL LINKS (in addition to above):
Diversity in Educational Software
http://webtools.cityu.edu.hk/news/newslett/Diversity.htm
Web Tutorials
http://www.techlearning.com/db_area/archives/
WCE/archives/tutorial.html
tutorials.com
http://www.tutorials.com/
elearnspace
http://www.elearnspace.org/index.htm
EdTechDev.
http://edtechdev.org/blog/
elearningpost
http://www.elearningpost.com/features/index.asp
http://www.keele.ac.uk/depts/cs/Stephen_Bostock/
docs/index.html
Highest Mountain, Inc. -- Compare Nokia to HP
http://www.highestmountain.net/showcase.htm

MORE:
The two competing theoretical camps are Constructivism and Cognition. David Jonassen is a leading thinker in the first group. William Winn writes cogently for a renewed emphasis on Cognition using an expanded definition.

Excerpts from:
A Manifesto for a Constructivist Approach to Technology in Higher Education.  David Jonassen, University of Colorado. 1992.

Generative processing involves relating new information to prior knowledge in order to build more elaborate knowledge structures. These knowledge structures are necessary for interpreting new information, reasoning from what is known, and for solving problems.

Too often, learning materials or environments are stripped of contextual relevance. Learners are required to acquire facts and rules that have no direct relevance or meaning to them, because they are not related to anything the learner is interested in or needs to know. . We believe that useful knowledge is that which can be transferred to new situations. The most effective learning contexts are those which are problem- or case-based, that immerse the learner in the situation requiring him or her to acquire skills or knowledge in order to solve the problem or manipulate the situation. So, instruction needs to provide contextually-based environments that are meaningful to the learners.

We believe that the goal of universities should be to produce reflective practitioners. … Reflection is a metacognitive strategy, that is, the process of thinking about thinking.

The point here is that browsing in a domain for which no properly developed schemata have yet been constructed is not likely to lead to satisfactory knowledge acquisition at all. Once someone has got to the point of enjoying the act of browsing in a domain, then learning from then on will be self-motivated. The challenge is to allow learners to get to that point. Construction tasks are needed first.

The most effective learning, that which is most meaningful and therefore transfers best, is case-based and involves meaningful real-world tasks. These are these experiences that contribute the most effective scripts to the development of expertise.

… we argue that observing other learners asking questions, receiving answers, arguing, choosing paths through the knowledge, or constructing their own knowledge structures by using a mind tool, is a fundamental learning experience. Indeed, one of the dangers of using technology, or learning at a distance, is that less of this kind of experience is typically available.
http://apu.gcal.ac.uk/clti/papers/TMPaper11.html

Jonassen interview. 2001:
Unfortunately, most e-learning replicates the worst features of face-to-face instruction. So, it may be cheaper to "deliver" knowledge over the Internet, but it will not be more effective. One of the requisites of substantive change in businesses and universities is to care enough about learning to invest the effort to truly understand its requirements and to create meaningful learning experiences to engage them. Unfortunately, that approach will be deemed as too expensive. You see, both businesses and universities operate from a bottom line perspective. And meaningful learning is simply too expensive.

I find it ironic that nearly every modern theory of management focuses on effectiveness, usability, addressing client needs, etc. Why does that policy never seem to affect training?
http://www.elearningpost.com/features/archives/001012.asp

Excerpts from:
Learning in Artificial Environments: Embodiment, Embeddedness and Dynamic Adaptation. William Winn. 2002.

It is disappointing that replacing the computational view of cognition with constructivism has led many educators to abandon a cognitive stance in any form. Ironically, this rejection is not found in the original constructivist writings.

Some of the same research, which led constructivists to reject computational explanations of learning, has evolved to a point where it can explain more complex aspects of learning while retaining scientific rigor, and remains centrally focused on cognition and learning.

The role of research is to find or develop functions that describe changes in the system, not to find systems that behave according to given functions. It is important to draw a distinction between "scientific AI" and "engineering AI" (Dietrich and Markman, 2000). The former is used to model and interpret how systems behave. The latter attempts to use established AI techniques to solve problems.

Mental, symbolic, representations of the world are real and necessary for cognition. Rather than "pictures in the head", these must now be thought of in terms of associative schemata, which have a neurological basis and which are activated by sensory inputs.

They argue that the way we organize ideas directly reflects how we act in the world. From there, Varela et al. construct a view of cognition that is based, not on the idea that the mind is a mirror of the environment, but that cognition consists of the constant, reciprocal, interaction between the mind and the environment. Even language is embodied. … We "have a hand in" a decision. We "look up to" someone. We "face" decisions. Thus, the very language we use to solve problems is often intimately tied to our bodies.

We have already stipulated that no-one's knowledge of the world can be complete, and that therefore everyone knows the world in a somewhat different way. But these differences in knowledge arise because everyone has a different set of experiences, not because there is no objective reality.

The experience of being coupled to an artificial environment is called "presence" (Zeltzer, 1991). Presence is the belief that you are "in" the artificial environment, not in the laboratory or classroom with a helmet on your head. Typically, during a visit to a virtual world, attention is divided between the environment created inside the helmet and the environment outside, which might be noisy, or contain someone giving you instructions about what to do, or be distracting in other ways. Presence varies with the extent to which attention is divided between the artificial and the real environment.

Our work with students of all ages has shown that presence consistently predicts the amount students learn and that reduced presence, caused by distraction or discomfort, impedes learning (Winn et al, 2001; Winn & Windschitl, 2002). These studies have also consistently shown positive correlations between presence and enjoyment.

Reyes & Zarama (1998) put forward an empirically-supported theory of learning, based on system-theoretic and cybernetic principles, in which they describe learning as a process for "embodying distinctions". The theory describes four steps. 1) "Declaring a break". … 2) "Drawing a distinction". … 3) "Grounding the distinction". … 4) "Embodying the distinction". … Reyes' and Zarama's view of learning parallels neurological accounts. If thoughts are represented in the brain as networks of neurons, defined principally by the pattern and strength of connections among them, then the brain is wired to make distinctions. The differences among neural networks can easily be computed from the differences among the strengths of individual connections.
http://depts.washington.edu/edtech/ticl.htm