Intelligent textbooks, rejected

I submitted a short paper to “intelligent textbook” researchers, arguing that we are still stuck in the 2-dimensional page paradigm, and offered a link to an interactive alternative in the references. As expected, they rejected the paper, and probably they did not notice how much their response confirmed my observation:

They wanted a diagram, instead.

(I deliberately did not include a static diagram because it is only after immersing for a few minutes in the interactive app that one appreciates the difference.)

Here is the paper


and here are the reviews:

----------------------- REVIEW 1 ---------------------
TITLE: Split Pane Interactivity
AUTHORS: Matthias Melcher

----------- Overall evaluation -----------
SCORE: -1 (weak reject)
----- TEXT:
Roughly the first two pages are irrelevant. The paper 
boils down to a claim that replacing pop-ups tied to 
the position of the item being referenced with the 
equivalent text being displayed in a fixed location 
is superior. This could have been explained far more 
clearly and succinctly without invoking the irrelevant 
issue of 2D interfaces being unchanged from paper. 
There is no empirical support for the claim, or even 
a significant argument that the claim should be true.

----------------------- REVIEW 2 ---------------------
TITLE: Split Pane Interactivity
AUTHORS: Matthias Melcher

----------- Overall evaluation -----------
SCORE: -2 (reject)
----- TEXT:
This opinion article advocates for alternatives to 
the conventional 2-dimensional page paradigm. 
Unfortunately, the article is hard to understand and 
I was not able to follow the main thesis:
* I am unfamiliar with split attention theory and 
the author doesn't define it
* I am unfamiliar with drone and ballistic design 
principles and the author doesn't define them

I recommend the author to work on the presentation 
of the main point(s).  Perhaps a diagram of what the 
author means by "split pane interactivity" could be 
helpful? Perhaps writing the paper in a more 
traditional "academic" structure could help readers 
understand the author's intent better?
The paper was rejected and the response unintendely confirmed
Posted in eLearning | Leave a comment

Large maps

After describing two other use cases of my think tool

I am going to describe a third type of maps: very large maps.

This use case is the one that is most similar to cartographic maps, and its purpose is to show many connector lines (‘roads’ and ‘streets’) and detailed items (‘villages’ and ‘towns’) but, at the same time, to retain a sense of orientation and gestalt. The orientation to see similar items in close proximity, and the gestalt to see recognizable patterns and clusters of similar items.

(Ideally, then, one would discover new connections within the condensed topic space. Or at least, become somewhat familiar with a new knowledge domain, by merely fiddling around and juggling a lot of its items.)

So, unlike many popular data visualizations, the map should not be an amorphous cloud of fancy arcs and tiny dots with random sampling just for marveling at, but rather, show identifiable connector lines and ‘palpable’, rearrangeable, items.

The orientation makes a difference that can best be seen in the short video here:

which is contrasting a navigation across an amorphous database, to a navigation based on spatial proximity respresenting the content similarity.

Now, the latest release of my tool offers some functions for teasing out how complex such a map might become before we are losing that sense of orientation. Max out the showable complexity by filtering more or less items and connections.

Originally, this third use case was not an intended application of the think tool. Rather, it emerged over time through single special maps as by-products. Click the examples below to see the respective posts.

Luhmann’s zettels

Folders (2)
Folders (1)
Words (2)
Words (1)
Forums posts
Posted in Tools | Tagged | Leave a comment

Sense-making workflow

In a previous post I showed screenshots of processing my own notes and ideas. Here, I want to describe how I use my tool trying to make sense of an existing text by someone else.

This example is the extreme opposite of the former: It was a long text, and because it was a difficult one for me, I used a “close reading” option to split it up into single sentences, and I did not add many thoughts of my own.

What you see below is the raw text pasted onto the canvas, each icon representing one sentence. Most icons’ labels don’t contain words yet. While reading one sentence after another in the detail pane on the right, I add a few important words to the label by selecting them in the right pane and pressing “B+” (Bold special). Similarly, I successively add the connector lines.

Raw unconnected itemsMost of the connector lines simply echo the sequence and hierarchy of sections and sentences. If the content of two distanced items is related, I use a different color (often yellow).

Connected iconsNext I recolored the connector lines.

Colored connector linresAnd finally i rearranged the map a bit. (Take a close look on the demo site.)

Rearranged final screenshotOf course, the map still looks dauntingly confusing, but I have learned a lot while making it. Note that I don’t normally use such a map to show it off to someone else.

Sometimes, I draw another map with reduced content. Many of my maps contain an entire paragraph per icon. Many contain additional items with my own comments, typically yellow icons simulating sticky post-its.

A few more details:

Sometimes, a paragraph does not contain content that is interesting enough to label it; then I re-color it into “pale”. Sometimes, the train of thought is only loosely connected, then I color the line “pale”. Some items represent headings, which I’ve come to recolor blue (because it is not too obtrusive). However, when I cluster some items around a “tag”, I often use red color for the tag, like it was a town on a geographical map which is surrounded by villages. The direction of the arrow lines does not really mean anything; often I just point them towards a major “town” just to preserve legibility.

Splitting up a text into paragraphs or sentences is not always satisfactory. In the above example, I found image captions or other inserts — but they can quickly be removed. Othertimes an abbreviation period or a running title disrupts a thought into two items — but they can quickly be connected. Conversely, when I think an item should be split, it is easy to copy and paste some part of it as separate item (Rightclick > Advanced > Paste here). I also used “Copy as list” a few times. For lazy labeling, I mostly use the existing words from the detail pane, even if the grammar of the label line then becomes awful. Rarely I type more text other than question marks, ampersands, commas etc.

Finally, I try not to forget to copy and paste the source URL (yielding a pale icon labeled “Source”), because the idea of thought-condensing only works when I can quickly get back to the uncondensed versions.

Posted in Tools | Tagged | Leave a comment

Remote work (updated)

Updated 2020-03-12: Of course it’s too late now to create a robust version of the synchronous whiteboard described below, too late for the current ‘Great Online-Learning Experiment’. I like CogDog’s advice to do “what the best teachers always do- improvise, change up on the fly when things change.” So maybe distribute and merge some copies of the map manually. See also Stephen Downes’s impressive list of tools, and his comment “We’ve had 25 years to be considered, careful, slow, and deliberate. Now it’s time to just start doing it.” (25 years)

Quick summary of a post and a video I made before the virus occasion.

Large walls of post-its are really very important to get a richer picture of a big project. But how does a large mural full of post-its fit on a screen? When you zoom-in so much that you can read the small print (as the participants standing at the mural can) the famous overview gets lost. But shouldn’t such a technical issue be solved by today? The video shows my proof of concept, using

Here, the details of an item are shown in a fixed corner. And despite what outdated doctrines still demand, they are not displayed close to the item, but where your eyes dart effortlessly, with ballistic saccades.Pictograms: three buildings connected across the globe

Posted in Tools | Tagged | Leave a comment

Notes workflow

In a previous post and video I described the workflow of my notes processing in an abbreviated sample demonstration. Now here are three screenshots of the authentic processing.

Note that this procedure applies only to my most precious notes which are not just related to a single project or topic. Still, of course, most of the daily stuff goes quickly to the project folders of the Operating System!

The grey icons below are the 61 precious notes I recorded since Nov 18, in raw chronological sequence, after revisiting them again, and after assigning them provisionally to some existing categories which are shown as green and black icons. The most interesting items are shown as red icons: these denote clusters of notes that might justify a new category.

Screenshot of stage1: note items are connected to categories but not yet rearranged

On the next screenshot below, the clusters are rearranged to facilitate another close examination, one cluster after another. Some cross references shown as yellow lines represent my doubts about the category assignment — and they may later be implemented as “see also” links/ pingbacks in my local WordPress.

Screenshot of stage 2: category revising in progress

On the final screenshot, the assignments are ready for the export to a WordPress WXR import file. One candidate for a new category (in red) has survived the process.

Screenshot of stage 3: ready for exporting to WordPressIt is probably obvious that all this effort is not just for quick retrieval of notes. If I do remember a keyword to search for, I can of course use the full text search of my WordPress. It is for browsing to find patterns that might not have a name yet.

Posted in Tools | Tagged | 1 Comment


Aaron Davis linked to an important article from 2017 by Tim Requarth of Slate Magazine about “closing the ‘information gap'” (via OLDaily).

“lectures from scientists built on the premise that they simply know more (even if it’s true) fail to convince this audience”

A gap in the pavement.

The article resonated with me in several respects:


“They may have more luck communicating if […] spending time on why it matters to the author”

This reminded me of the problems that many have with learning the style of scientific papers. Everybody agrees that such papers have to be emotionless and sober, and everybody complains that they are boring and tiring. Of course, the author must not employ emotional rhetoric to try and sway the reader into believing an unsupported claim and compensate for its weak evidence. But reading would be much easier if the author emphasized why some aspects are more important to them, and why they are passionate to find a convincing solution.

But — as if they were not able to distinguish these two kinds of emotions — most authors cultivate a special style that not only sounds maximally unbiased, impartial and emotionless, but also escapes into weird, rare and distanced words, and eventually conveys the impression of disinterest and disrespect and even arrogance towards the reader. I acknowledge that developing such a habit is not easy, and once acquired, this style becomes the hallmark of scholarly ‘speak’ that shines through in each and every public utterance. Certainly, some scientists may be so frustrated about the academic atmosphere of envy and elbow rule that they have indeed become blunted and emotionless, doing without passion whatever the funding agencies currently want to hear.


“communicators can be more effective after they’ve gained the audience’s trust. “

This is the most important, and seems obvious to me: trust has been lost, and the gap has become too wide.


“it may be more worthwhile to figure out how to talk about science with people they already know, through, say, local and community interactions, than it is to try to publish explainers on national news sites”

This reminds me of Downes’s successful networks, which are not centralized hub and spoke networks, but those that work through a ripple effect and propagate the messages on trusted shorter pathways.

In a polarized environment, we cannot expect that someone abruptly changes their mind against all of the closest acquaintances, without gradually seeing their neighborhood rethink, too.

Posted in Learning | Leave a comment

Snake oil serendipity?

James Stuber reminded us that

“A good carpenter doesn’t use a swiss army knife; they have a toolbox. Each tool has a job to do, and each tool does that job well.”

and I replied:

“Exactly. In particular, Swiss Army Knife tools suck when they try to cover filing, meaning making, and collaboration. My promises only the second.”

Since then, I have been reminded of how different the jobs are that are done even with note management tools.

While the workflow I described in my new video is about just a few important notes, other usage scenarios involve a huge mass of pages:

A graph of 435 densely connected dots, colored by clusters.
This map depicts a roam user’s database (whose export file was freely downloadable), or more precisely, the 435 well-connected pages out of 1759 total pages. Obviously, no layout can visualize any reasonable gestalt here, in particular in the blue cluster in the lower right.

The motivation of such hoarding of links for later is obviously the hope for generating “future serendipitous connections”.

Is this hope justified? I feel a bit guilty that I may have suggested something similar myself: In a previous blogpost “Magic of Zettelkasten” I tried to explain the uniqueness and the surprising success of Niklas Luhmann’s approach with this very serendipity, as he found it in his huge collection of notes.

(And I almost regret having used the term ‘magic’ which attracts plenty of clicks but may sound like snake-oil promises which are currently spreading. Furthermore, what is currently being discussed as ‘zettlekasting’ does not have much in common with Luhmann’s original method, except the ‘atomicity’ of the individual ideas noted on separate index cards. For a more original version, see Daniel Luedecke’s tool rather than

The difference, however, is that Luhmann did the organisation of his cross-references manually, and I think this is what helped him to find serendipitous insights later. I doubt that the raw mass of machine-generated full-text matches can have a similar effect. When I played with the above-shown data, I was overwhelmed with tons of “Unlinked References”, and checking them one after another, will probably be more numbing than inspiring. I know how much work it is to sift through the keyword concordances of a corpus linguistics tool like Antconc, but at least these keywords are sorted by relevance and they are promising. Maybe the next step will be to employ some AI to fulfill the sprawling desire of getting one’s outstanding ideas as effortlessly as possible — Devonthink comes to mind.

An alternative approach for getting connected ideas would be, IMHO, to be open for them directly while reading a text. However, the goals for note-taking about one’s reading are very different, and some goals seem incompatible with this openness.

When notes are taken for a literature review (as in the above-shown example), an understandable strategy is to do this mass effort as economically as possible. Maybe to the extent that the “goal is to never have to return to the full text” (source). This, IMHO, enforces a style that is very much convergent, and focussed on the text at hand rather than open for one’s own divergent associations, and it forces into wrapping and fixing and into ultimate and definitive wording rather than gradually evolving understanding.

(In fact, I am rather proud that my own tool enables just the radical opposite of this style, in that you can use provisional summarizing labels just because you can always go back quickly to the details. And as I understand McGilchrist, the wrapping and fixing and focussing is just the opposite of a broad vigilant open attention that would enable new associations to pop up.)

Similarly, when the goal is to learn, in turbo style, things to remember, one’s efforts are being directed to passively absorbing the text at hand rather than associating new ideas. A telling indicator is for me, when I see students in the train with multiple marker pens but no pencil to write margin notes or questions or disagreements. I doubt that all the ideas come later when re-reading the marked stuff. For me, reading works best when I am permanently at the ready for deciding to write (voluntarily) about what I have just learned, e.g. to blog about it.

Posted in Tools | Leave a comment

Automagical links?

(I realise this has become a series, with part 2 here and part 1 here, about which kind of tools do help (me) thinking and which do not.)

Recently, there has been much buzz, even a “#cult”, about a new think tool that did not impress me at all.

A pedestal with a golden engraving saying "Cult".

What does it offer that seems so fascinating to some enthusiasts? It offers hyperlinks, backlinks, and autolinks (“unlinked references”). OK, hyperlinks are not very uncommon in online texts; backlinks are not very exciting if the links are also strokes on a map. And automatically generated links — I have more hesitation than fascination for them.

In my experience, automatic links tend to clutter up my display, both in text views and map views. Often they link things together that only incidentally share a term, and as in most search results, the ‘hits’ are often a miss. Often they are redundant and just add the transitive composition of two existent links.

So with this tool, my workflow would need another step, where I sift through the matches, deciding if these links are valuable serendipity, or clutter.

But I don’t like to sift through. With the information tsunami of the online world, this sifting habit has become a big ugly part of our lives. Yes, we are getting better at that and faster (to the extent that our reading habits are altered). But it is straining and makes tired. My guess is, this is because the workflow sequence is externally prescribed by the list from the machine, rather than by our own self-determination and discretion.

(I don’t like linear lists, generally. Probably for the above reason, that my mind feels somehow patronized and channeled with ‘leading-strings’. The long indented bullet lists of the hyped tool, are another feature that puts me off.)

At first sight, of course, the serendipitous matches catch one’s eye, and they probably cause a fascination for the results that are automatically (automagically) generated. And “networked thought” is a great idea (I use that term myself on my experimental quotation maps and pages). But once a tool is so autonomous that my own thinking process is interfered with, I react old-fashioned and become a Luddite.

So maybe I can generalize this sentiment: I don’t want the tool to present me with many artificial creations, or to dominate our joint efforts, but to augment my own way of doing it. I want a multi-purpose tool like a ‘hammer’ that fits closely in my hand, rather than an autonomous machine that spits out the ‘deus ex machina’, or generates hundreds of artificial thoughts that I need to sift through. If the tool’s strength is coping with the multitude of ideas, it can serve me well if it just sorts that multitude — such that browsing the list feels more intuitive and less straining.

Think tools should not consume my attention with their autonomous creations, but help me sorting things out for myself.

Posted in Personal Productivity | 1 Comment

Inspiring or distracting?

Gradually I can tell better which kind of tools do help (me) thinking and which do not.

In a previous post I contrasted the ‘sovereign-posture’ app with the ‘clerk’ that Engelbart envisioned, i.e., the tool that helped augmenting the human intellect. Apps that push themselves into the foreground, seem more patronizing than helpful to me.

Now there are also applications that come with plenty of endowment (provisions/ equipment/ ‘furnishings’?), and some fittings suggest themselves as proposals/ prototypes (templates/ ‘sand moulds’?) for our thought contents.

Eight colorful sand noulds lying in the sand

Image background adapted from Sand patterns in the dunes CC-BY-NC by Flickr user jennymackness

One vendor describes this in their video as follows:

“[The app] looks over your shoulder and offers to help” (here at 10:01)

Maybe for some users, such offers may function as an idea generator or a creative muse. For others (like me) they do not work; for users who know distraction-free software, such offers will even appear as obstruction and patronization.

Typically, software won’t accommodate both sorts of users. So, instead of just going ahead, I am often required to find or create a type (sand mould?) for my stuff. This classification, however, is exactly the paralyzing problem of premature pigeon-holing that prevents to see the gradually emerging connections — which is the main goal of our Condensr.

Posted in Personal Productivity | 2 Comments


Stephen Downes’s talk “about a way to redefine ethics” contains a lot that sounds plausible to me. In particular, I liked this:

“we learn ethics, but we learn them in such a way that we feel or experience a moral sense, rather than fully formed general principles” (slide 70 76)

I know almost nothing about philosophical ethics, so this is just my learning aloud. The simple 100-year-old distinction between “the ethic of attitude and the ethic of responsibility (‘Gesinnungsethik vs. Verantwortungsethik’)” by Max Weber, 1919, won’t suffice when the latter relates to Consequentialism/ utilitarianism/ hedonism/ liberalism, and the former to Deontology, rules and rights and perhaps virtues.

Map clipping of Max Weber's house

Image: Max Weber’s house in Heidelberg, on map by © OpenStreetMap contributors

Anyway, I have often been surprised by how many people today expect that some fixed scientific ethics should give them certainty about what rules to follow, and that it could perhaps somehow exonerate them from the accountability of choosing their own principles. As if ethics was mandatory like the law and not a voluntary contribution beyond the law.

This week, the talk pointed me to the ethics of care and its awareness for vulnerability and dependence. While vulnerability and dependence in education are no doubt a familar aspect for teachers or parents of smaller children, I must confess that I had not spent much thought about them. After all, in higher education, gaining intellectual independence seemed to me a crucial goal because, simply, someone must be able to decide even if there is nobody available who can be asked!

Now the “care perspective”, in the immediate situation, seems to take on the responsibility without trying to ‘delegate’ it. But the decisions and criteria need not always be independently derived. They may be learned from others who were in a similar situation and whom we may ask about how they would decide here. We may also ask others about what ethics they follow, but the ultimate accountability of adopting their ethics for ourselves, cannot be offloaded on to these others, or on to some higher authority.

Learning this ethics takes different ‘contagion’ paths, so to speak, which vary with the decreasing dependence. For infants, the cognitive ‘navel string’ is from mother and parents, later from family and friends, colleagues and communities of practice — the path is the same as for the primordial trust to be seeded and then grown. This percolation path may not yield perfect results and may be slow to change. But it is robust against nonsense from a central, influential source — just as Downes’s “successful networks” promise.

Posted in Learning | Tagged | Leave a comment