Are daily keystrokes a good proxy metric for overall productivity? How can I measure them?

Question

Lately I've been getting interested in "quantified self" metrics. I've been tracking basic health metrics like weight, sleep, caffeine/alcohol consumption, and several other subjective health and mood ratings over time for a while now. The data nerd in me likes trying to tease out the relationships in this data to eke out a little more productivity each day. When the gathering and analysis time are factored in I may not gain much productivity, but I still enjoy measuring my progress and seeing how the numbers plot.

I'm looking for great proxy metrics for productivity.

The problem I'm running into is that abstract notions like "productivity" are more difficult to quantify in a meaningful way. I am seeking proxy metrics that are simple, objective, and unobtrusive.

Both passive timetracking (RescueTime) and active timetracking (Beeminder/TagTime) miss the point as far as I'm concerned. I tried both for a while before realizing that time spent working is not a proxy for intellectual productivity. Measuring time spent working is a sure path to maximizing how much I'm staying busy, which is not a goal of mine.

Another proxy measure I've experimented with is to track "pages of technical material read per day." This data is interesting to know for its own sake, but as a proxy for overall daily productivity it has a few flaws:

1. Doesn't reflect the whole picture: Technical reading (and associated notetaking) is only one small part of my job. I alot ~1-2 hours per day to this task, but it doesn't measure my effectiveness in other areas.

2. Distorts my priorities: Using this as my productivity metric causes me to prioritize it over other tasks, even when reading another technical paper is not my Most Important Task for the day and I might be better off working on other things.

3. Not objective: "Pages" itself is a subjective measure. A page in Science is not the same as a page in Nature or any other journal. All these subjective questions come up that have to be handled each time. How are figures counted? What about math? Do I deserve any kind of bonus if the paper was especially arduous? I've developed rules of thumb that I generally follow for these, but I don't like that my metric is so subjective.

4. Recording error: It's sometimes inconvenient to record the measurements, since reading dense technical papers is something I prefer to do with hardcopies for easier highlighting. After reading, perhaps later that day, I like to sit down at the computer and write up a summary of what I learned and what I think about a particular paper, but by then my recollection of how much I read is a bit fuzzy. If I were more vigilant this wouldn't be a source of error, but if I'm being honest I must concede that I am probably not 100% accurate in recording pages read after the fact.

Am I being a producer or a consumer?

It occurs to me that perhaps I am asking the wrong question. Maybe I shouldn't be asking "Was I productive today?" Maybe to get quantitative I need to drill down and get a better definition of productivity.

For me, being highly productive means producing information of value. What do I do when I am producing? Well, as a knowledge worker that usually means typing. I could be typing emails, or notes, or papers, or comments, documentation, code... any number of things.

Contrast this with my favorite timewasters, which are mostly passive: reading news, stack exchange, checking facebook, etc etc. All these activities are passive (except for the occasional overlong question or answer on SE).

This dichotomy in my typing intensity suggests a very simple, precise, objective, unobtrusive measure of daily productivity!

Maybe you're all wondering: Is there a question in this question? There is!

Do you have any experience tracking daily keys typed as a proxy for productivity?

I'd really like to hear about your experiences and any potential flaws in this as a measure of productivity for the average knowledge worker. But most of all I want to know....

What tools have you used to track this information?

This is really the heart of what I'm asking since "key tracking" or similar google queries seem fraught with danger. I don't want to install keylogging malware in my quest for efficiency, so if you know of any reputable open source projects that offer typing statistics I'd be very glad to know about them.

Answer 1

I don't think keystrokes are a good metric. You can sit and type and backspace all day long to get a high keystroke count, and not have any productivity. Or play typing speed games.

You don't want to measure activity, you want to measure results.

I suggest a better proxy is "pages produced", using some standard definition of a page. For example, page preview everything to a standard sized page and note the number of pages when you complete a task. Track the number of pages per day as your productivity proxy.

Another option is simply number of tasks completed. Yes, tasks are of different sizes, but over time that will even out and a simple number of average completions per day over a week will be a pretty good proxy.

One of the things I've learned about using proxies to measure any behavior is that you want to look at them over a period of time, not a fine level of granularity. If the question is X per day, you really want to look at a week at a time to see trends, not absolute numbers day by day.

Answer 2

I don't think keystrokes are a good metric for productivity unless you are a typist.

Having read through this, I don't think you even know how to define productive activities and unproductive activities.

How productive you are is a balance between what you have to achieve and the time you are taking to achieve it.

Answer 3

To come back to this from a slightly different angle than previous answers - I'd genuninely be interested in having a utility that monitored by keystrode rhymn and pace - it would interest me to know what my typing speed was for various activities, such as writting emails or even code - and for what activities typing speed really is the limiting factor for my work. It would be good to know how accurate normal 'typing speed' tests stand up to this more imersive approach.

Anyway- an answer - following this question I've been playing with http://code.google.com/p/osxkeylogger/ - (obviously only of use if you are on OSX) which gives out logs of the form

<key><time>352983060774062</time><type>10</type><keycode>13</keycode></key>
<key><time>352983164758362</time><type>11</type><keycode>13</keycode></key>
<key><time>352983300902040</time><type>10</type><keycode>4</keycode></key>
<key><time>352983364607367</time><type>11</type><keycode>4</keycode></key>
<key><time>352983404743412</time><type>10</type><keycode>0</keycode></key>
<key><time>352983492739069</time><type>10</type><keycode>17</keycode></key>
<key><time>352983533134639</time><type>11</type><keycode>0</keycode></key>
<key><time>352983566845481</time><type>11</type><keycode>17</keycode></key>
<key><time>352983612688474</time><type>10</type><keycode>39</keycode></key>

which should be pretty simple to convert into keystokes per day, ect - I'm going to look into using them to see answer questions like 'what is my average words per minute when really writing?' and 'Can I eliminate my use of the mouse from some programs by using the shortcuts'... unfortunately it doesn't appear to work in terminal though so that will hurt pretty bad...