Indexing, context and understanding

Inspiration

Smarter than crows' Meg Mcgowan wrote a really nice blog post based on a review of Lynne Kelly's book, The Memory Code.

(You can find that post here: Books that change your brain: Lynne Kellys "The Memory Code")

I've wanted to write a article based on some of the ideas used in both the "Memory Code" and the follow up guide "Memory Craft" for a while now, and Meg's article reminded me of that. For myself the focus was slightly different. Not so much about patterns, but of "indexes" and "indexing", because it's one way of thinking about the tools evinced in these two books.

I should say that, in a nutshell, the books are about memory techniques that you can use to improve memory retrieval. These same memory techniques allow users to store truckloads of data, and more importantly, easily retrieve it.

Using your brain
instead of your smart phone

Where we rely on smart phones and computers to store (and index) information, non-literate cultures (and this includes classical Greek and Roman orators as well as indigenous peoples of the world) as well as modern day memory champions use their brain and a technique Lynne refers to as the method of loci, to store and index information and easily retrieve it.

If you are looking for "brain training" exercises, look no further.

As mentioned, my focus in this article is on indexing, and why that is important across a variety of realms. (How does indexing relate to memory techniques? I'll talk about that also!)

I'll also attempt to show how good indexing can be a measure of understanding or at the very least, lead to better understanding. And I'm going to talk a little about holons, because it's a concept or "idea" that I've thought alot about and written about but using a different name. (I tend to think of holons as "ideas"). And it can relate to the idea of indexing and the idea of context.

A Holon, something that is both a part and a whole

In the article, Meg refers to a holon thusly: "a holon in systems thinking is essentially the boundary that you decide to place around something in order to study it, while keeping in mind that all things are connected. "

I studied Systems Design engineering at university, and while we were exposed to the idea of drawing boundaries around a system in order to study it, I don't think the word holon was ever used.

Note the wikipedia given definition of a holon: "something that is simultaneously both a whole and a part."

When I was in university we were taught that we could define a system in more or less the same way, by drawing a boundary around it. The thing inside the boundary was the system. That which was outside it was the environment. One of the important ideas was that we could choose how we drew the boundary.

The important thing, whether you call what is inside the boundary a system or a holon, is that it is clearly defined.

More on this, later.

Memory palaces as a type of "indexing system"

When memory practitioners use a memory palace or some physical device (all based in some way on the method of loci) to store information so that it is easy to retrieve, what they are creating (at least one way to look at it) is an indexing system.

The memory palace is a collection of organized (or "related") indexes that can be used to point to a specific memory.

Memory palaces and other memory tools make it easy to pull up information quickly. And using lots of memory palaces can make it easy to store lots of information and still pull it up quickly.

Juggernauts of indexing

As to why indexing is important, we can start of by looking at two modern day behemouths, which rely significantly on effective indexing: Google and Amazon.

Note these two behemoths use indexing in different ways, but I'm going to talk about just two examples, one from each.

Google indexes web pages based on the ideas that they talk about. A single page from a website, can be listed under a multitude of indexes. Any of these indexes can be used to reach the particular page (or any other pages that use that particular index.)

Meanwhile the company Amazon uses indexing in its warehouses to allow it to get products out the door faster.

In Amazon warehouses, products are stored wherever there is room. (And that's sort of like data on older hard drives, divided up, or fragmented so that it can be fit in where ever there is room.)

Because products are stored where there is room, an up-to-date databased is required.

To make finding a particular product easier, Amazon indexes all the locations of a particular product (and presumably the quantity of said product in each location) so that it can send workers to the nearest location (or locations) that contains that product in the required quantity.

So while Google uses indexing to help you find a web page, Amazon uses it so that it can send workers quickly to retrieve a particular product.

In either case, indexing makes retrieval quick and easy. Time isn't wasted figuring out where the "thing" is. The location is known. In once case the location is an address on the world wide web. In the other, it's a physical location inside a warehouse.

Providing context for memories

If you ever notice the "keys" that bring up a particular memory, it seems that the brain uses some sort of indexing system.

(At least that's the way my brain seems to bring up memories, and I'm assuming that there's at least a few people who have brains that work in the same way as mine.)

Memories are linked or indexed via shared qualities, whether it is a particular emotion, a fear of something, or some shared quality that the brain can identify and thus link with other memories that share that quality.

As a side note, context is important.

I'll see someone out of context and think that I should know that person. Passing someone on the street the other day, it was only after I passed him that I realized he was one of the guys who works at a breakfast shop I infrequently frequent.

And this can be one of the limitations of relying on our built-in memory systems. When deliberately trying to extract information, we have to know the context the memory is stored in.

With devices like "memory palaces", the palaces themselves provide the context.

Creating a framework for memories

An important point with regards to the use of memory palaces is that there is some set up involved, prior to using them.

In The Memory Code, Lynne introduces the many different memory devices she uses on a daily basis to store different types of information, whether the information is the various monarchs of England, the bird species of Australia, the countries of the world or even historical figures she finds meaningful. In Memory Craft she goes into a lot more detail on the various devices, including some short term devices that can be used for temporary storage for things like grocery lists.

From personal experience, the two biggest challenges I had with using tools like memory palaces is building them (or in the case of physical locations, choosing them), and maintaining them.

As a side note, I've also made the mistake of trying to use as memory palaces, things that weren't varied enough in appearance and/or not very interesting.

The nice thing with memory palaces is that you have some control over your indexing system. You can design it. You can modify it. Knowing the context, the memory device, you can then pull out memories indexed in that palace or storage location at will.

Without these add on systems we have to rely on our "natural" memory. And again, this can often be slow if the particular piece of data is limited to a particular context. Or if the context isn't obvious.

Landscapes, architecture and physical objects as memory devices

For indigenous peoples, memories where initially tied into the landscape. Lynne's big realization, covered in detail in the Memory code, is that Structures like Stonehenge arose when indigenous people become more settled. These people-made artifacts were away of capturing the essences of the various landscapes a people would regularly visit.

These "memory places" enabled the keeping alive of memories as well as enabling them to being passed on to the next generation of memory keepers.

In some cases, memory devices were hand held. You would tour the landscape of these devices by touch (and perhaps by sight as well). The device was a framework for holding or indexing memories.

As the device and the memories it was linked to became self contained, i.e. memorized, the physical device was no longer needed. Both the device and the memories it linked to had both been memorized. As a result, the device could be discarded. The memory of it, and all that it linked to, was all that was required.

This is an important point, and perhaps easily glossed over. The memory device itself is memorized, as well as the data to which it points to. In a way this is like google (but in reverse.) Google was build to index the web. But it's now become a very part of the thing that it indexes.

The memory palace, or memory device serves as a road or path linking all of the memory locations contained by that device. You could actually think of memory devices along with their information as systems. Or if you like, neural networks that we custom build.

Memorizing, the old fashioned way

My own understanding and experience of memory and memorization has arisen in a few ways.

Plays and promises

I've spent my fair share of hours memorizing lines for plays I've been in. I also spent time memorizing obligations for the lower three levels of masonry. I'd recite the obligation on my way to and from classes, carrying my little book with me so that I could check to see that I'd got it right.

In either case, the idea was to be able to speak my part without thinking about what came next. With an obligation, I was simply repeating a string of words. Acting in a play I was repeating a string of words in response to whatever was happening in the scene.

Poems and phrases

Even before I moved to Taiwan, I spent time memorizing phrases of Chinese characters so that I could paint them from memory.

One of my first attempts in this regard involved learning the relatively short and simple first verse of the Dao De Ching. I would paint the verse from memory, while in the flow. The resulting piece would reflect my state of being at the time of painting.

In general, paintings that I painted while I felt calm, peaceful, content, would reflect the same feeling on the person viewing said art.

Later I tried memorizing the 1000 character essay so that I could paint or write it from memory.

This is literally an essay with 1000 unique Chinese characters. It is broken down naturally into phrases of 8 characters each, so all I had to do was memorize a set of 8 characters at a time, then it's place within the whole.

Note that in both cases, my focus was on the characters and a rough idea of their meaning. I didn't really learn their pronunciation. Instead, I memorized the characters of these verses so that I could paint them freely while focusing on the experience of painting.

Sequences of movements for Tai Ji and Yoga

I've also memorized various tai ji forms and the ashtanga yoga primary series.

(In both cases there are people who can only do these forms by "following along".)

Learning Ashtanga, I had the benefit of a book that included all of the poses. I would simply memorize three or four poses at a time, adding new poses each time I practiced, and checking the book after the practice to see if I'd gotten them right.

With Tai Ji, I've done the same thing, watching sections of a youtube video enough times to learn a few moves so that I can practice them before repeating the process for the next set of moves.

Whether doing yoga or tai ji, if I'd memorized the sequence of poses or postures, I could do those without thinking. Instead I could focus on more important things like feeling my body and controlling it.

Holy holons batman, those definitions really bring out the details

The most important thing I can say about memorizing, particularly when not using special memory techniques as outlined in either of Lynne's books, is for whatever you are trying to learn, break it down into clearly defined smaller elements. Then learn the whole as the sum of its parts.

Here the idea of holons is important. You can break one big thing down into many little things, each a thing in its own right, but still a of the bigger thing. More importantly, each little thing is recognizable as a thing in its own right.

Another idea is to choose the level at which you break things down. Where possible, you can choose this so that your break down is as flexible as possible.

In any case, the important idea when breaking things down is to make the break, the definition "clearly defined".

That's what a holon is, a clearly defined part of a bigger thing.

As a side note, anther trick when memorizing is to look for similiarities. As an example, with the primary series of yoga poses from the Ashtanga system, I looked at things like the distance between feet in various standing poses to make the sequence of standing poses easier to remember.

With Chinese characters, sometimes a particular stroke of one character might be noteworthy enough that it serves as a hook for remembering the next character.

Taking a stroll down memory lane

I should mention here that I've had other experiences of memorization that relate to "understanding".

I've memorized the components of various weapons that I had to work on in the army, electronic systems I worked on after university, and more lately a rough mapping of the interconnection of bones, muscles and connective tissue in my own body.

In each case what I memorized wasn't just the parts, but how the parts related.

And in a way this is very similiar to how Lynne describes the process of memorization using small memory devices.

She might take the dog for a walk through various neighborhoods around her house, taking a trip through history as she does so because various locations along one particular route she's used to index periods of history. Familiar enough with the route, she doesn't actually need to travel the route in order to revisit all of those places. It is built into her memory.

In a similiar way the process of learning how a particular rifle worked first involved taking the rifle apart and putting it back together again repeatedly. Thus we learned all of the pieces. These pieces would be the equivalent of the individual periods of history Lynne ties to various locations along her walk.

Once we were familiar with the pieces, we then learned a sequence of operations that occurred during the various modes of operation. We'd literally take a journey through the weapon noting how each part affected the next part in the chain of operations.

The end result was that I knew the rifle inside and out, and with some extra instruction on top of that, eventually learned how to fault find and fix rifles. I had a memorized version of the rifle, of it's parts and how they work together built into my memory.

This same process of understanding occurred when I was working with a telehealth system (basically a primitive version of facetime or skype, purpose build so that doctors could have access to patients in remote locations).

I learned the various components of the system, say for instance the camera, the mic, the video card, the audio card, the codec (which stood either for compression and decompression or coding and decoding). Knowing the components, I could then look at specific signals one by one and trace their journey through the system component by component.

I could, for example, trace a video signal from the camera as it travelled through the system, first to the video card and from there to the local display (where it could be displayed side by side with feed from the other end) but also to the codec and from there to the modem where it was sent via ISDN line to the other location. At the other location it would go from the modem through to the codec to be decoded. Then it would travel to the video card and from there sent be sent to the computer screen where it was displayed with that systems local video.

I learned the system well enough that I could talk clients through fault finding procedures without me having to resort to a manual, I knew the system that well.

I even learned the system well enough that when we had to make a change based on a customer request, I was able to come up with a solution that was cheap and easy to implement in a seeming flash of inspiration where all other suggestions, made by people who hadn't learned the system as well, involved the purchase of expensive additional equipment.

At the time I had never heard of memory palaces or the method of loci, but I was more or less using the same technique. But instead of memorizing information, I was understanding systems.

Moving to Taiwan

I first began learning about the importance of indexing, when I moved here to Taiwan.

On moving, one of my first self set tasks was learning the various Chinese characters for food and drink. Here it was important for me to be able to say the word as well as read it (so that I could both recognize what I wanted as well as order what I wanted). To that end, I'd end up in various coffee shops or restaurants, writing out characters and looking them up in my tiny hand held dictionary one-by-one.

My dictionary had a radical index with characters sorted by radical. It also had a stroke count index. And it had a phonetic alphabet where characters where listed according to their pronunciation. This I couldn't use because I didn't know the pronounciation of the characters I was looking up.

• The radical index was frustrating to use because there was always a character whose radical was hard to figure out.
• And the stroke count index was even worse. Did that shape count as one stroke or 2 (or 4)?

Traditional Chinese character radical and stroke count indexes are less than ideal

What I eventually realized as I began a process of building my own database of Chinese characters, (said database growing to include meanings, pronounciation and eventually shape based typing codes), was that Traditional Chinese character indexes are less than good for looking up characters, especially when you are a beginner.

Of all of the available indexes, stroke count was the worst. Did a particular shape count as one stroke or two? As mentioned I couldn't use the phonetic index because in most cases I didn't know the pronouncation of the character. As a matter of fact, one of the reasons I was looking up characters was to find out their pronounciation. So the radical index was my main recourse to Chinese character lookup.

And while a character's radical can help you figure out the meaning of the character, it's not very useful when you are actually trying to look the character up.

And it doesn't help that the positioning of the radical varies from character to character.

And so what I ended up doing, slowly, was first realizing that I needed an index system for my database, and then second, creating one.

An alphabetic index system for Chinese characters

The first indexing system for Chinese characters that I came about using was pre-designed as a shape-based input system. It took me a few years to realize that the shape based typing codes I was using to type Chinese characters was actually an alphabetical shape-based index system for Chinese characters.

With cangjie input codes, the codes allow the computers operating system lookup a character and output it on the screen. I eventually realized that this same system could be used to "alphabeticize" Chinese characters by shape.

And so I was excited when I finally realized that these typing codes were an effective indexing system for Chinese characters. While there were some inconsistencies, in general it was a lot less frustrating to use than traditional Chinese character indexing methods.

Improving the index system for Chinese characters

While I did use this as a "sensible" indexing system for Chinese characters, after a short while I thought that for beginners it would be a bit of a pain because they would have to learn how 24 letters of the latin alphabet matched to corresponding character elements. And in general I was finding that people tended to avoid memorization as much as possible.

And so I reduced the indexing system down to 12 shapes with 16 sub shapes.

How was this different than say using a radical index to look up characters?

The element you were using to look a character up was always in the same position relative to the character. It was either left most or top most (or in a few cases, the outermost element).

Later on, I developed a second index based on the second element of the character. I did this because some groups were quite large, and while they could easily be subdivided, I thought a good way around this was to provide an index where you could lookup characters based on their second component.

There were some issues in creating this index though.

• How do you define the second element of a Chinese character? While this seems straight forwards there are characters which can be divided in a number of ways, and depending on your experience with Chinese characters, the placement of the dividing line can make a lot of difference.
• Am I designing for those familiar with Chinese characters and thus biased towards stroke order or for those that aren't.
• And what about for people already used to the cangjie input system?

I bring these problems up because when figuring out how to index something, thought is often required. But something else to bear in mind is that indexing can always be changed or varied (ideally in such a way that it is improved.)

Later I also added an index based on the final or "third" element which in turn was keyed to the final stroke (the right most or bottom most element of a character).

And so what I ended up with was three different shape-based indexes for looking up Chinese characters.

What is the purpose of an index

At this point I should stop for a moment and point out that an important idea to ask about indexes is "what is the purpose of this index?"

When the radical index for Chinese characters was first created it was perhaps with the mindset of grouping characters by meaning and not so much to make them easy to lookup.

For myself, the idea in creating Chinese character indexes was to make Chinese characters easy to lookup for anybody.

And this is very much the idea of memory devices.

Multi-purposing indexes

A secondary consideration for myself with respect to Chinese character indexes, but also important, was making the indexes useful for learning certain aspects of particular characters, whether it was their meaning, their pronunciation or even to remember particular elements.

As an example, I found that with rear element sorting in particular, I could subgroup characters via their pronounciation. And this then could potentially make it easier to remember that characters with a particular shape would have a particular pronounciation. This wasn't true in all cases, but it was true in enough cases to take the time to additionally subsort characters in this index via their pronounciation.

(As a side note, I did also create a phonetic index, and additionally, a rhyming index. This latter one was handy because often characters with pronounciations that rhyme will share similiar elements.)

Learning to read Chinese by Reading It

Note that doing all of this indexing didn't help my own Chinese. At least not initially. What I was doing was creating a tool to make it easier for me to lookup Chinese characters. That then made it easier for me to learn how to read Chinese characters. How am I doing that? By reading the translated version of Stephen Kings It. Actually, I started with that. Now I'm reading his book "Doctor Sleep".

I say "read". It's more of a crawl. I'll read a section in Chinese, then the corresponding section in the original.

Translations tend to be sentence by sentence and paragraph to paragraph. As a result, reading the English, I can get the gist of a sentence or paragraph, and then lookup individual characters easily based on their shape easily using my app.

Interestingly, this forces me to read the English with more detail.

To help me memorize the different characters I'll use a mix of the techniques.

In some cases I'll us the techniques learned from a few months with the Mandarin Blue print method. (This uses variations of the techniques Lynne talks about in her book).

In other cases I'll look at the same character in different indexes to see it in different contexts.

Does its sound have any relationship to other characters with the same elements and sound? If not, what other characters that I know of that have that sound can I match it to?

With reading, spaced repetition is built in

The nice thing with trying to learn by reading a novel is that spaced repetition is built in. (In addition, the novel I'm reading is one I know I like, so the interest is there also. )

(Did I mention that an important part of using memory devices is reviewing the information?)

However, even with the built in spaced repetition that is provided in a novel, I did decide to build a history function in my app (it didn't have one to begin with). So now I can also look at words that I've just looked up. So whatever memory device I've used for a particular character, the history function lets me review it quickly so that it makes it past the short term memory barrier.

Becoming your own teacher

By indexing Chinese characters (even if I didn't know their meaning) in an external device (I've made an app using the index, and I've also made pdf's and ebooks using the same lookup system), I made it easier for myself to lookup Chinese characters on the fly. Then, having learned various memory methods including the Mandarin Blue print method, I'm now taking the time to index Chinese characters internally, to the point that eventually I won't need the app that I created.

In either case, indexing is key. It's allowing me to teach myself Chinese through the simple expedient of reading it.

Yoga poses and a yoga pose website

I developed a further understanding of indexing both through yoga and through building a website about yoga poses.

Should tree pose be classified as an externally rotated hip pose?

As an example of how indexing can be used improperly, A long time ago I was on a yoga teacher training. There was some discussion as to whether or not tree pose should be considered an external hip rotation pose or not. The conflict arising from the idea that in all other external hip rotation poses the externally rotate leg was bearing weight.

I thought at the time it was a stupid idea and a stupid argument. The categorization wasn't important to the doing of the pose. Or perhaps more importantly, it was possible to categorize (categorize = index) a pose in different ways.

For the record, I would count tree pose as an externally rotated hip pose. But I'd also categorize it as a single leg balance pose. I'd also count it as a sort of "binding pose" because when you press the lifted leg foot against the standing leg thigh you change the forces acting on both legs.

Perhaps the point that was missed here is that it is possible to categorize things in many different ways. Going back to the idea of holons, you can choose how you draw the boundary that defines the holon (or the system.)

Indexing a website about yoga poses

As for indexing poses (and pages of yoga poses) that's something I'm now trying to do more effectively in my own yoga website.

In earlier renditions of my website, I was a slave to the idea of SEO and the idea that each page should be linked to a single specific idea or indexing term. While this was important I was neglecting something more important. Making it easy for people to find stuff on my website using site navigation (and not just relying on Google.)

As a case in point, I found it difficult to find articles on my own website. And I'm the one who build the website. And I've written all the articles that are on it.

My indexing was less than ideal.

And so now I'm in the process of indexing pages based on their multiple relevances.

So for example the transverse abdominis. It works on the lower back. It also affects the ribcage. It also affects the SI joints. And it directly affects tension in the other abdominal muscles, like the sliders used on tent lines in the old days. And of course it can also work in concert with the diaphragm.

Then there's the sartorius. It works on the knee and the hip. And it's potentially affected by adductor and vastus medialis tension. And while it can bend the knee, it can also externally rotate the thigh while at the same time internally rotating the shin. And it can be a key player in alleviating some types of knee pain.

Why am I talking about muscles, when the website is about yoga? Because if you understand muscles, and how they work,then you can better use your body (since muscles not only move your body, they allow you to feel it).

So now what I'm trying to do is allow both google and any organic visitor to my website easily find articles with effective indexing for both parties. And now, rather than having to figure out how I've indexed a particular page, ideally any page is easy to find using the web sites own navigation, based on any of the main ideas that page talks about.

How does this relate to the idea of context? Basically, I'm creating multiple contexts for looking up each page on my site based on the information it contains.

Indexing makes retrieval easier

Earlier I mentioned that me indexing Chinese characters hasn't actually improved my Chinese. However, having built an app (and other devices for looking up Chinese characters using the same indexing) I can now use my app to look up characters easily. And that does help me learn Chinese.

In general terms, indexing (and appropriate contexts) removes the road blocks caused when trying to find stuff.

• Google makes it easy to find what we are looking for on the web because they've indexed the world wide web.
• Amazon makes it easy for their workers to find product in their warehouses because they index product locations.
• With Chinese characters indexed intelligently, I can now spend time actually learning characters instead of wasting time looking them up.
• In the same way, if I make it easy for visitors to find stuff on my yoga website, they can get on with learning instead of struggling to find stuff.

Context is important too

An important part of indexing is context. Sometimes the context is so obvious we miss it.

When you use google you'll get a mass of search results. Those search results are all indvidual indexes within the context of your search term (or terms).

With Amazon, the context could be thought of as the warehouse and the location of a particular product within the warehouse.

With Chinese character indexing, each type of index can be thought of as a particular context. In once case it is characters sorted by first element, in other cases by second or final element. And yet in other cases it is characters grouped by the initial part of their pronunciation or their rhyme.

For a website about, ahem, yoga poses (and stuff related to yoga poses, i.e. bones, muscles, joints, the body), the context can be poses, or it can be stretching or strengthening or parts of the body etc.

Is the index for one person or many?

Something that should be pointed out is that sometimes a set of indexes is designed for many people to use. That's the case with the indexing on my website. And it's also the case, potentially, with the indexing system I designed for Chinese character lookup. In cases were an index is being designed for a wide range of people, it can help to have various contexts so that the things that are indexed are easy to find by as wide a range as people as possible.

When dealing with memory for our own use, a single index for each thing stored is adequate so long as the context that contains it is clearly defined.

Indexing for better understanding

In the case of understanding, particularly when we want to share that understanding, it can not only be helpful to index the parts of the system in different ways. It can help to break the system down in different ways.

Using the yoga website as an example, postures themselves can be grouped according to type of position, sitting or standing etc, or whether they are can be used to stretch or strengthen or both. Looking at the body, we can look at it from the view point of joints, and the muscles that work on each one, or we can look at it from the view point of individual muscles and either the bones or the joints that they affect... or both.

I'll suggest here, the more ways you can break down a system in different ways, and index those parts, and relate them, the better your understanding of that system.

Good indexing allows you to get on with what you are doing

Effective indexing makes it easy to get on with what you are trying to do. Instead of struggling to find information or things, you can get on with using that information or thing or sending it where ever it is needed.

Using indexing and context as a basis for understanding

Perhaps one of the most important things about indexes and their context is that they are the basis of how we understand things. As Lynne mentions, in her book, the method of loci involves storing specifici pieces of information in specific locations. With the method as used by indigenous people in particular, the information was often related to its storage location. This is as opposed to simply using a memory palace or other memory device for simple storage of "random" information.

The relationship between the device and the information that was stored in each place was meaningful and had an actual relationship. So in a particular stopping place, refreshing a memorized song or dance or combination of both that was related to that particular place, all kinds of information could be brought up, but in particular pieces of information important for the survival of the group in that particular place such as where water was to be found and how best to access it.

Learning to understand various systems to the point that I could explain them, fix them, fault find them etc, we more or less used the same sort of techniques. But they weren't known as memory techniques or the method of loci, they were simply the process of understanding how a particular system worked. The system itself provided the physical location, with each component part a potential location. Meanwhile we could traverse the system while tracing various signals of sequence of changes through the system, once signal or one change at a time.

And it's pretty much as Lynne describes one particular aspect of memorizing, particularly while using hand held memory devices, but also when the memory devices are physical locations. Once learned, you don't need the actual device. It's in your memory, your consciousness, where you can travel through it at will.