First some questions of clarification:
Could you [Hannifin] please post the references to McCaslin & Good; Richard Clark; and to Bob Kozma. [I agree the style of listing influential experiences is important and much missed in conventional academic literature; but it makes "conventional" reference lists more, not less, important.]
I wasn't always sure of how the "revelations" were intended to relate to associated text. In particular: Revelation 3: did you think animal behaviorist studies were "researching how people learn" or "researching if people can learn in particular ways?" Revelation 4: explaining versus predicting. Explaining is a vague word, and I often have to ask people to explain what they mean by "explain." In particular in this context, two contrasting meanings occur to me. In a study we might show learners an apparatus where a ball is dropped down a helical tube and ask them (a) to predict in what direction it will travel when it shoots out of the lower end and (b) to explain their prediction, i.e., to justify it. In this sense, explanation is asking something more than prediction. Another meaning however might apply to my amateur theories of the effect of position in a family, such that when I ask someone if they were the eldest child I then often feel that explains some of their personality traits--explain in the sense of relating my observations to a framework. But this sense of explanation is much weaker, because actually it is not able to predict behavior more than very weakly. It amounts really to a classification of new events in terms of old ones. It may be completely empty (like astrology), or it may be the forerunner of better understanding, as when physicians learn to group symptoms into syndromes before they later gain understanding of causes, predictions of outcomes, and treatments to avert those outcomes. In all cases, it is a different sense of explanation, and weaker than prediction. Which, if either, did you mean, when you said that explanation had become more important to you than prediction?
And now a comment or two, as you requested.
COMMENT 1: The paper immediately made me think about the issue of when "instruction" is worthwhile. I have bought a book to help me learn HTML, so I can get more able on the web. I slightly resent having to pay, but it feels definitely worth it: it shortens my learning time. This is a good case, because for anyone with some computer training, the web is a good self-teach by exploration environment. There are all sorts of materials available, you can look at the source of any web page for examples, and you can experiment very very easily by trying out your own web pages. This might count as an OELE. Yet the book, very old fashioned teaching material, seems a clear, though modest, value for money. I feel this the more strongly as I have up to now assumed that teaching myself was better for me; but the book is illuminating all sorts of details, clearing things up. So on the one hand it can stand for why teachers are worth paying, even in a very short sighted sense (i.e., no need to wait for lifetime effects, imponderable values: the payoff is immediate and clear to me). On the other hand it might not meet the definition of instruction mentioned in the paper (no assessment, no paper wasted writing out learning objectives). It is also interesting in that I of course use the book to pick out what I want, to set myself activities. It is BOTH an old fashioned piece of instruction AND actually supports my own tailoring to my needs, as would any book that is adequate for reference purposes and adequate as a linear narrative for me to read/skip through to get a basic grasp of what the topic contains without having to have fully formed questions in advance.
Perhaps this example isn't truly either instruction or open-ended learning; but it is real, and it made me aware of the value of "teaching" even of very moderate quality in speeding my learning.
COMMENT 2: I think the issue is echoed in another domain with a time scale orders of magnitude smaller: the design of user interfaces. It is approximately true to say that you cannot use a human-computer interface without learning how to use it, although today a typical time scale would be seconds or minutes, rather than the days and weeks for learning an academic topic. And designers face the issue of whether to support users learning by exploration (cf. OELE) or learning through instruction (cf. the minimalist instruction technique established by Jack Carroll).
And this domain also throws up an issue sometimes debated by educationalists. If learning by exploration (LBE) is supported well enough, users end up never learning: they just re-experiment each time. This doesn't matter with user interfaces, though it often surprises designers. But educationalists worry: they worry if pocket calculators mean children don't learn mental arithmetic, if pictures in books mean children never learn to understand the text without that prop.
COMMENT 3: I am currently interested in how this issue has appeared in teaching programming to students. It used to be the case that you didn't have to have debugging in the curriculum, because you couldn't learn to get programs running without also picking up a lot of debugging skill. In the last few years I and others have noticed that our students cannot debug any more, although they write good and well structured programs. This is because current programming environments are so supportive that students eliminate bugs rapidly by reacting to the error messages. If you give them a listing, they often cannot even recognize correct vs. incorrect syntax because they have never needed this skill. Finding a bug by reading the code is beyond them.
Their programming environments are OELEs that are very supportive. This is probably most valuable in the early weeks. As a teaching aid, though, they are perhaps a liability that seems to prevent learning debugging (perhaps the main transferable skill). Perhaps it is a simple lesson: we have to progressively withdraw the support. But that is not being done, not least because supportive environments are now normal: and will be found in work environments. We may have to make the OELE less supportive, and less ecologically valid, in order to give them enough practice at debugging.
That is one of my experiences of technology and OELEs, and it seems to require the disabling of the OELE in favor of a more directive instructional approach (as part of the teaching) in order to equip students with a vital skill that will not be required with sufficient frequency without this artificial intervention.
Of course you might say that the compiler and associated environment was not designed for education specifically, and you might possibly say that my book on HTML wasn't designed either: or at least not by the educational/instructional design methodologies the paper referred to. But this brings us back to the observations early in the paper about how people often really learn, independent of the favored learning theories or design methods. Nevertheless, the same issues seem to emerge, one of them being directive instruction vs. other supports for learning. Another, illustrated by the HTML case, is that superficially directive materials may nevertheless support a more autonomous approach by the learner. So a question for OELEs perhaps is: is the converse true? Can they also support a more directive learning style? Good modern textbooks show the influence of instructional ideas (e.g., explicit objectives, built in (self) assessment, etc.). But they also take care not to lose their functionality as reference books (index, contents, bibliography, etc.) and so can be used by learners who entirely ignore the instructional structure and use them as resources for other modes of learning. How do OELEs shape up? Do they all need to have appended "wizards" or "agents," or appendices with sequenced lists of learning objectives to act as directive guides to those who require that approach?