Way, way back in February of this year I was trying to improve the voice card interface and was frustrated because I needed more hard data about your reading habits. How long do you spend reading cards? Do you read straight through or skip around? That kind of thing.

The same kind of problem is faced by engineers designing missiles and spacecraft. How do you find out what's going on inside your gadget when your gadget is hundreds of miles away? The answer is Telemetry. Have your gadget monitor itself and broadcast the results back to command central.

So that's what I did. Inside Voices Volume Four I installed a series of routines that silently transmitted data about your reading activities. Every time you opened and then exited a new card, I recorded the time to the second. By subtracting the two values I could determine the A.R.T. (Apparent Reading Time). I say "Apparent" because I had no way of knowing what you were actually doing while the card was on the screen. You could have been answering a phone call or passed out cold. But it most cases it's a safe bet that you were reading.

I also recorded the time and the exact location of the mouse click whenever you left the voice card stack (to go to a previous volume or whatever). I measured how long you spent (apparently) writing cards, how often you returned to the index card, and even how long your computer had been turned on before you entered the voice volume.

And since issue four included my Mr. Wizard program "Cube of Crete," I also recorded your movements through the maze down to the sixtieth of a second. I had intended to release all of this information in the following issue, but I got distracted by a certain woman, as you know, and, well, better late than never.

WHAT IT IS

The telemetry program is a very simple stack that does little more than simply store the data pulled from your transit stacks. Six-and-a-half members sent data (Roger had technical difficulties and Holly sent only partial data). The data in its raw state is an almost unreadable stream of numbers, so I translated it into a more readable format before storing it.

Each of the seven member cards contains a large scrolling field which can display either the raw or translated data. Since each card is given only its short name in the field (e.g. "VC 4 Holly 12") I added a pop-up index field so that I could quickly find the full title of each card (clicking on the title jumps to the original card).

At the end of each field I added some totals. For each member I show Total Apparent Reading Time, Total cards read (out of a possible 103), and Average time per card. I also show total time spent inside the voices 4 stack. This is not the same as Total A.R.T. because the per card reading times were only recorded for the FIRST time the card was examined, and some cards were examined more than once.

I defined a "session" as turning on your computer, fiddling araound for a while, and turning it off again. Most of you logged about four sessions over a period of a few weeks. For each session I recorded the total time from the moment the Voices 4 stack was first entered to the last recorded moment it was exited. Finally, I showed a grand total from all the sessions. Although any sessions in which the voice stack was not touched would not be recorded, I feel this grand total is probably pretty close to the total time spent dealing with Archipelago in a typical issue (with 103 cards).

The results? Most of you average from about 30 seconds to a minute per card. The total time spent inside Voices 4 was an hour or two, and the grand total ran about four or five hours. During issue four Tomás was working on his Diabetes stack and managed to rack up a record ten hours. There was quite a bit of bouncing back and forth to the index card, and several of you were unable to read all the cards before sending your reply.

As for the Cube of Crete: I stored enough data to completely re-create your movements through the maze, and by feeding that data into my maze program I was able to watch five of you move through the maze just as if I was looking over your shoulder. I added your total moves and total elapsed times to the end of the main data field. The results were as follows:

Cliff	4 moves in 1 min 24.5 secs
Stuart	8 moves in 1 min 40.6 secs
Larry	14 moves in 2 mins 38.2 secs
Paul	49 moves in 8 mins 1.1 secs
Robert	41 moves in 12 min 55.4 secs

As a special treat I haved added a visual display to the front of the Telemetry stack. On the opening card you will see an array of 103 member icon buttons representing the 103 cards in volume 4. At the bottom of the screen are icons for the seven participating members.

When you push one of the seven member buttons you are given a choice between examining the raw data and seeing the display. If you choose to see the display, the program will clear the screen (if necessary) and then light up the icon buttons IN THE SAME ORDER the corresponding cards were read by that member. After each button it will pause very briefly for a time that is roughly proportional to the time spent reading the card. In this way you can tell which cards were skimmed and which cards attracted more attention (although the actual length of the card is not factored in). If you click on one of the icon buttons the full title of that card will be displayed on the screen.

The display is lots of fun and gives an immediate overview of the rather massive data for each member. It is also worth while to explore the data itself in more detail. An incredible amount of information can be deduced from a close inspection of dates and times. It's interesting to see different reading strategies in action and to evaluate the efficiency of each apporach. As with all stacks in this series, there is a Mr. Wizard button on the top card so that you can bounce back and forth between the program and this screen. Have a look!

My analysis of the data led to several new features which I hope will make the reading faster and easier for all of you. Since issue four I have not gathered any telemetry data, and I will not do so again until I get your reaction to this idea.

There are some exciting possibilities. Larry has often used reading times in the course of his psychology research; perhaps if he is interested and we are willing, we could rig up some kind of reading experiment. Also, the telemetry concept can be expanded into other areas and could be used, for example, to re-create races and other real time games.

Please look through the telemetry stack and let me know what you think. Remember to scroll to the bottom of each data field to see the totals. And, as always, feel free to look inside the button and card scripts to see how they work. Enjoy!