Thursday, December 04, 2014

Borges's Taxonomy

My speculation on taxonomy has driven me back to an old chestnut. It's easily found online but I figured that I would repeat it here (courtesy of Wikipedia).

Borges gives us a kaleidoscope of authority. His taxonomy comes from an essay about John Wilkins -- founder of the Royal Society -- and his efforts at creating a universal language. Borges then cites a passage from the mysterious and ancient Chinese encyclopedia called the Celestial Emporium of Benevolent Knowledge, giving us a culturally relevant and appropriate taxonomy of animals:

  • Those that belong to the emperor
  • Embalmed ones
  • Those that are trained
  • Suckling pigs
  • Mermaids (or Sirens)
  • Fabulous ones
  • Stray dogs
  • Those that are included in this classification
  • Those that tremble as if they were mad
  • Innumerable ones
  • Those drawn with a very fine camel hair brush
  • Et cetera
  • Those that have just broken the flower vase
  • Those that, at a distance, resemble flies

The list is very informative in that it we see all of the challenges of developing a taxonomy. It represents functional driven analysis run amok, co-mingled with classification by ownership ("belong to the emperor") and short term objectives ("just broken the flower vase"). Some are perhaps driven by economic necessity ("suckling pigs") while others must be part of some sort of speculative project that belongs in the pages of a China Mieville novel ("tremble as if they were mad").

My particular favorites are "et cetera" -- primarily because it appears in the middle of the list -- and the self referential "those that are included in this classification", thereby undermining the orthogonality of the whole taxonomy.

Wonderful stuff.

Training 2014/11/28 #001

It's not really the first training session, but is the first recording of one...

Escape from standing hammerlock

Position: standing; defense; back-to-opponent

I've always been impressed with SlideyFoot's willingness to document his training and teaching sessions. It's something we should all emulate.

Last week, I learned something new. It's an escape from the basic standing hammerlock (i.e., the basic cop "you're coming with me" move).

The defense is relatively basic. Maintain posture; make some space by sliding out the foot on the non-entagled side; throw the non-entagled elbow at your opponents head. You're basically throwing the elbow behind you. There are a few fine points here. The first is that you're moving your hips in addition to your body (kind of like tai no henko ni in Yoshinkan Aikido). The purpose of this elbow throw isn't so much to make contact but to make some space. Think of it kind of like a reverse Sambo-style casting punch.

The purpose of the elbow is really to get your opponent into a guilotine choke. Again, with the guillotine in place, square your hips; focus on driving the elbow through; get a gable grip. Then you have two options:

1. Takedown. Sprawl and take them to their knees. Ideally, you want to push your partner down on an angle (i.e., through your centre-line). When they're down, you have some options: hold them, move to side control, whatever.

2. Pull guard. Pull guard and keep control of their head. To get the submission, think about hyperextending their neck by extending your hips, etc. Remember to squeeze your knees to keep control.

UPDATE -- we subsequently explored what to do if you end up with the arm-in guillotine. Sneak your arm through, grab the inside of your opposite elbow, talk on the phone, and squeeze. The other option is to grab the cloth at the armpit, and then wind up your body to get the choke.


Of course, you might be on the receiving end of this move. What do you do if you find yourself in a guillotine?  The first step is to tripod up to make some space to breath. Hold their knee with your offside hand and then pass via a backstep or knee slice to move to side control. Your head is still trapped but you're okay. Again, how you have options. You can apply an aggressive crossface and choke them on their own soldier. You might be able push down the wrist of their entangling arm to get the kimura. Or you can free your head while maintaining control of the wrist to set up a standard kimura, etc.

Escape from the escort headlock

Position: standing; defense; back-to-opponent

I'm really not sure what this position is called. Basically, you find yourself in a headlock and getting dragged backward. Your opponent has your neck and you have no posture, balance, or position.

The first step is to ensure some air supply. Grap their entangling arm with both your hands, one on either side of the elbow. Turn away from the choke. Basically, you want to try to get your trachea in line with the crook of their elbow. If you turn the wrong way you basically wind the choke tighter.

The second step is to get some posture. Don't pull against their arm. Instead, drop and rotate. Basically, you're swinging your bum out. At this point, you've probably broken their posture some.

The third step is to block the back of their inside foot with yours and then slide your other foot forward. As their balance shifts backwards, they'll likely fall. Come to think of it, the foot change is a bit like tai no henko ichi. Overall, the throw is a bit like a very mellow osoto gari.

As your partner goes down you still have control of their entangling arm so you can move to an arm press, a kimura, etc.

Escape from punching bully headlock

Position: standing; defense; inline-with-opponent

You might find yourself in a standing headlock where your opponent is punching you with their other hand. Think of it as the druken brawl headlock.

Step 1 is to defend yourself from the most immediate threat: the incoming fist. Reach around their back with your inside hand and grab the elbow of their punching arm. Don't let them punch you in the face!

Step 2 is to improve your posture. Grab their entangling arm to make some space and tuck your tail to create better posture. Look up. Think "rhino horn"!

Step 3 is to then move the entangling arm so that their forearm is vertical and then push up. The force vector should no be applied at right angles to their upper arm. If possible, get your rhino horn into their pit and force them off balance or on to their toes.

Step 4 is to step back a bit and get the cop hammerlock. You might even want to get the crossface chickenwing!

Escape from the standard bully headlock

Position: standing; defense; inline-with-opponent

You know this one if you have a big brother. You're standing, your opponent has a headlock on you, and they have broken your posture.

The defense starts in a similar manner to the punching headlock but the finish is different. Turn slightly to the inside so that you get some air.Reach around their waist with your inside hand and either get a hook or grab some clothing. Put your outside hand on your side of their inside knee. This hand just stops them from coming into your space. Stretch your inside leg behind their outside leg and then drop. Think of dropping your butt immediately on to your outside heel.

Ideally, your opponent will go down backwards, hard and you'll come around into side control. Isolate an arm; get the submission.

Labels: ,

Anthony Bourdain on BJJ

Yeah, I do BJJ. People often ask: why? It's always a bit unclear if they are asking me why I do it or why I am passionate about it. The answer to "why do it?" is pretty straight forward: good workout, community, self defense, etc. But defining why I'm passionate is a bit more difficult. Fortunately, celebrity chef Anthony Bourdain has basically provided an answer:

Tuesday, December 02, 2014

Organizing the workstation

Enough with file plans and records series. Let's get back to actual users. I'll start with a 2007 article from InformationR and co-authored by an old classmate: Brendan Luyt (

The paper explores how people put files on the hard disk of their workstations. The results indicate that people typically use on to three levels of folders and that the top level folders tends to organized by task or project. Browsing folders was more common than searches. Folder names often reflected document type, organizational structure, and the great catch-alls of "misc[ellaneous]" and "temp[orary]."

The literature review describes a 1989 study by Kwasnik that suggest seven dimenions of information organization:

  • situation attributes (source, use, circumstance, access)
  • document attributes (author, topic, form)
  • disposition (discard, keep, postpone)
  • order/scheme (group, separate, arrange)
  • time (continuation, duration, currency)
  • value (importance, interest, confidentiality)
  • cognitive state (don't know, want to remember)

Situation attributes, particularly use, see to be most common.

A mathematical model by Balter (2000) suggest that -- for email -- users should have between four and 20 folders to optimize searching, and that using more than 30 folders is inefficient. It should be noted that this argument applies primarily to active folders and there may be a larger number of infrequently accessed archive folders.

So, we're getting some cohesion with functional records management. Use tasks, particularly at the top levels. Limit the number of folders. Depend on search for granularity. So this gets us to a few other questions: how do we do the analysis of tasks and how to do we deal with the whole subject issue. Tagging?

Monday, December 01, 2014

"Modern filing"

It's time to go back to a classic: Modern filing : a textbook on office systems. Yes, I used quotation marks in the title to indicate irony. Apologies.

The book was printed in 1916 and published by the Yamwan and Erbe manfuacturing company. It was basically a sales catalogue. To put things in perspective, in 1916 a filing cabinet was still a relatively new-fangled thing, Scientific Management was near its apogee, and the United States had not yet entered the Great War. Ypres and the Somme had perhaps already distracted Canadians from recognizing the importance of modern filing... but that's a different discussion.

The table of contents gives us an overview of various systems: the loose sheet, the Shannon file, vertical filing, numerical filing, direct name system, geographical filing, subject filing, etc. The introduction is paean to the wonders of improved office management and gives us some insight into the conditions of the day:

"In many large business organizations there are chief filing clerks who have from two to thirty assistants who do nothing but look up and file correspondence. There are thousands of business houses throughout the country that require from eight to fifty people to take care of their filing. Expert librarians have been hired by business houses to supervise filing at salaries running as high as $50.00 a week. So you can see the subject is worthy of the utmost attention." (p. iix)

We no longer have this level of commitment to filing but we have the same problems:

"Frequently it will be found that business houses are struggling with filing methods they have far outgrown, or that are not adpated to their requirements. The results are not what they would like, yet they cannot see any  way to improve the methods they are using." (p. ix)

In fairness, the introduction alludes to other problems that are completely unintelligible to modern readers -- the difficulties of putting correspondence into a "transfer case" or negotiating with printers over the need to change card stock sizes.

The first chapter is about the "loose sheet system". It started with a basic box to contain letters. It evolved to cabinets of drawers where one could essentially put stacks of paper. The problem with this system is that some drawers fill up more quickly than others, requiring the use of aforementioned "transfer cases". These cases then go into longer term storage. We then end up with a retrieval issue since some drawers will contain years of correspondence while others will have only a few months with older correspondence in the transfer cases.

We move on to the Shannon File, which "like the Loose Sheet, is termed a Small Drawer system to distinguish it from the Vertical or Large Drawer method." A Shannon file really seems to be a clipboard type of arrangement with a manilla index or overlay. The key innovation seems to be the tabs that give rapid access to various letters of the alphabet. There are also Shannon Drawers, which are essentially clipboards that can be stored in a cabinet. The advantage of the Shannon system is that pages can be lost from the top rings. There is still, of course, the transfer case problem. NOTE: the form factor of the transfer case seems to live on as those boxes that store current journals or magazines in libraries before they are bound.

Then we get to vertical filing, or what we would call just filing. The system seems largely unchanged in almost 100 years and the book's illustrations are immediately recognizable. We are told that things are filed chronologically, most recent on top. A name-based filing system seems to be the default classification approach.

The book then launches into "Methods of Indexing and Alphabetical Filing". No surprisingly, we see the same list of approaches: alphabetical, by name; geographical; and subject. And of course, a hybrid approach (or in the terged text of the day: "Sometimes it is found convenient to use all of them."). In alphabetical filing we also have the transfer issue, etc.

Two things fascinate me about this description of filing. The scope of the paper problem is comparatively miniscule. Managers apparently worried about the need to scale up from four-drawer vertical files! The other issue is the nature of the business. Everything seems to so personal. The assumption is that everything can be tracked by name. It is all about correspondence between individuals. How times have changed...

The book goes on to reference the Numerical System and notes that it is the oldest form of vertical filing (perhaps a hat tip to Dewey's Library Bureau?). The numerical system makes certain things easier (like cross indexing) but introduces the need for some kind of index and surrogate. In this manner we can provision folders in a FIFO manner or by accession number and then manage reference through the surrogate. It's basically the classic library problem of optimizing both the user's ability to find a book and the library's need to shelve them!

One version of the Numerical System is the Direct Number system with Dewey-esque guides for the 10s, 100s, etc. This system still echoes in functional classifications with three primary levels of granularity.

Concerns with numerical systems include the general clunkiness of doing lookups, the waste of folders (because every concern needs a folder), etc. We are told that for "best results" we should separate correspondance by years and provide capacity for two years before transfer. That's probably good advice that could apply to any filing system such as email, etc.

The system "rapidly taking the place of other methods" is the Direct name system... unfortunately, the description is a bit opaque. Basically, you maintain the numbering system but also list the name on the folders and then separate frequently used folders (which might have monthly subfolders).

It's interesting that a manilla folder is optimized to hold no more than 50 pages of paper... could a similar threshold exist for email folders?

The Location Method holds few surprises: manage correspondence by state; when you exceed a particular threshold, use cities; etc. What is interesting is the implication of this system. The amount of correspondence indicates the size of the potential market in that area! It's like proto-BI. Of course, we then get into the mysteries of transfer tracking, etc. which really no longer seem relevant.

Now, on to Subject Filing, that subject that seems to give all of us so much grief. Thankfully, the text notes that to "index a subject file so that its contents are accesible is not generally difficult." (p.43) It continues by noting that subject filing for correspondence is only used in large companies but that it is most appropriat e for managing data files... which, at the time, was apparently a pretty rare thing. It also, however, notes that subject filing in purchasing may be valuable for documents related to current projects but that for permanent storage an alphabetical or numerical approach should be applied.

And that's it... nuts. We don't really get a lot of guidance on how to actually create a subject index!

Next, we get special cases such as the "Follow-Up" file, also known as a:

  • Tickler
  • Jogger
  • Calendar
  • Every Day File
  • Pending File
  • Dates Ahead File

There are variations on the approach but most people will be familiar with the concept from either their document management system or the Getting Things Done productivity technique.

We then get into some fascinating discussion of how to copy paper via letter press, rapid rollers, carbon paper, etc.  Thank goodness for technology!

Chapter XI introduces a "Card Record System": "The business man who is his own manager, book-keeper and office boy uses card systems." (p.59) The cards can serve as an index but they can also serve as a record for things like quotations, real estate records, etc. We then start to see odd work-arounds like the use of coloured metal tabs to use for follow-up, etc.

The book then gets into some mysterious details of how to deal with things like mortgages and legal documents, the use of cards for managing accounts and inventory, etc. Finally, we get the stamped due date record. The book was borrowed once, in January 1933.

Have I learned anything from this? Yes. Filing alphabetically is important. Folders are important.

Categorization, classification, oh my...

It seems I'm a bit old school. When discussing overly pedantic (but occassionally necessary) guidelines I'll mention AACR2. It's crucial for libraries but a bit overkill for most business users.

It seems that AACR is winding down in favour of Resource Description and Access (RDA). Like any good standard, RDA is not without its critics and opponents. What's interesting, however, is why RDA is necessary at all. Why did AACR fall short?

A key enabler for RDA is Functional Requirements for Bibliographic Records (FRBR). Specifically, FRBR articlates that an Item is an _exemplar of_ a Manifestation, is an _embodiment of_ an Expression, is a _realization of_ a Work.

Hmmm... interesting and all but probably not that useful for what I want to do.

The IFLA Statement of International Cataloguing Principles gives us some additional insight ( Specifically, it notes that "bibliographic and authority data" can include the following entities:

  • Work
  • Expression
  • Manifestation
  • Item
  • Person
  • Family
  • Corporate Body
  • Concept
  • Object
  • Event
  • Place