Failsafe

President Obama recently proposed an intriguing solution to deal with his country's ever-growing debt: a "failsafe" trigger. Here's roughly how it would work: if the debt as a share of the economy (the debt-to-GDP ratio) does not drop below a certain ratio by a certain date, then spending cuts, tax increases, or both would automatically be implemented.

Politically, it's a brilliant solution, as it transfer the onus of decision-making from the politicians to the bureaucrats, as they are forced to make the deep but necessary cuts to lower the debt, currently a nightmarish $14 trillion.

A failsafe system can be applied to any process. It is comprised of:

1. a specific, measurable goal to be achieved
2. a date or time period by which the goal must be met
3. a specific, measurable action that will be taken if the goal is not met

A simple example is weight loss. For example, you could set a goal of losing 3 kg. in two weeks. If you fail to meet this goal, you would have to lower your intake by 300 calories, exercise an additional 30 minutes, or both. You would repeat this failsafe system until you have reached your ideal weight. Then comes the tough part: maintenance. A second failsafe system ensures you stay on track.

Could a failsafe system be developed for documentation? It could, if we can define a measurable goal. However, objectively measuring the quality of documentation can be difficult. To obtain an objective, measurable goal requires carefully observing a user interact with the documentation.

Some of the measurements of documentation could include:

• the success rate at which the user finds the relevant topic
• the length of time required for the user to find the relevant topic
• the average rating given a topic by the users

Another measure could be the number of contacts (phone calls or emails) to technical support. This could be broken down further into:

• contacts made due to incomplete or inaccurate documentation - for example, a procedure is missing or a field is not explained clearly
• contacts to report specific documentation problems - this occurs when users give direct feedback on the documentation

After deciding what it is you are going to measure, you can then set a goal based on a specific date or time period, for example:

• improving the success rate by 15% over the next 3 months
• reducing the topic search time by 20% over 6 weeks
• reducing technical support contacts by 10% by August 1

Finally, you need to select an "action item". That is, what specific action will be taken if the goal is not met?

Ideally, the action would be be implementing a thorough review of the documentation (or portions of it) based on feedback from users or internal staff such as QA, Business Analysts, Product Managers, and so on.  This could include creating a "closed feedback loop" whereby users can directly comment on any topic; the results are then sent to the appropriate writer who will make the necessary changes.

Is this idealistic? Yes. Many technical communicators are already stretched to the limit, so asking them to set aside time to improve their existing documentation is not always realistic. However, for those who are fortunate enough to implement such a failsafe system, the end result is documentation set that actually saves a company money.

Read More

The Chinese Tent

Welcome to the Chinese Tent...

Imagine a large tent, and in it, a person who speaks only English. We'll call this person the respondent. Outside the tent is a person who speaks only Chinese. We'll call this person the questioner. The questioner writes a message in Chinese and passes it to the respondent through an opening at the front of tent. The respondent receives the message, then using an infinitely complex look-up system, matches the message on the card with a corresponding card containing a response in Chinese. The respondent delivers the pre-written response back to the questioner.

From the perspective of the Chinese questioner, the English respondent has read the message and understood it, demonstrating intelligence. However, has the respondent actually responded intelligently? Remember, the respondent does not understand a word of Chinese - he has simply matched up a pre-defined response to the received message.

This is not intelligence, but an elaborate forgery. The Chinese Tent exposes the folly of thinking that computers can actually be intelligent. Computers can simulate all sorts of reality; this does not make the reality they simulate real.

When a computer runs a virtual stock market exercise and makes millions of virtual dollars, the user does not become wealthy. When a computer runs a weather simulation that has 40cm of rain falling in one day, the user doesn't get wet. Yet when a computer simulates intelligence, we all rush to believe it's actual intelligence, and not a pale imitation.

It all comes down to definitions, something technical communicators are very fussy about. So how do you define intelligence? Here's one standard dictionary definition:

"the capacity for learning, reasoning, understanding, and similar forms of mental activity; aptitude in grasping truths, relationships, facts, meanings"

The key word here is understanding, which implies consciousness; the knowledge that one exists. This is something that computers simply don't have. 

However, if you define intelligence as the ability to give a meaningful response to a question, then computers could indeed be intelligent. A recent example of this was Watson, a super-computer developed by IBM. Watson was a contestant on the Jeopardy game show, beat his human opponents and won a million dollars. 

Depending on the definition of intelligence you choose, Watson is intelligent or simply another version of the Chinese Tent. It is able to analyze questions and supply responses, but if you were to ask it how it felt about winning against humans, it would have no response.

What, then, is an "intelligent" document? It is one which understands the information requirements of the user and then seamlessly delivers them. Now, there are help systems that allow a question to be entered and then try to give a relevant response, but these are very complex to set up and have mixed results. A far simpler solution is to create a complete and meaningful index, one which anticipates all the strange and wonderful ways a user might look up a topic.

An index also has the advantage of revealing potential gaps in your content or the index itself.  For example, you may have the following index entries:

records,

  creating

  editing

  deleting

files,

  creating

  deleting

Where is the entry for editing files? Doh!

 
A complete and meaningful index is an intelligent index, and helps to make your document intelligent. Of course, once a user finds the topic they are after, it has to be intelligently written. This means that the topic should answer questions, and not raise them. Questions such as:

• What is this object, thing or concept?
• What is it used for? Why would I use it?
• How do I perform a specific task?
• What are specific things I need to be aware of?
• What things or tasks are related?

An effective document, then, is one giant answer book. An intelligent document gives the user the answers they need, without the user even realizing the effort they made to find them.

Read More