December 18th, 2022
Raw Data
I embedded into the code for LMD a routine that would save data on each move made by each player. Here’s a typical line from the resulting file:
Erin Wed, 14 Dec 2022 17:00:21 GMT 834 1112 ---297,1; 1.070 0.180 0.870 0.170
The first word, “Erin”, is the name assigned to that player. I refuse to compromise the privacy of my players, so when a new player first appears, the program assigns them a random name and then uses that name to identify that player henceforth.
Next comes the date and time that the move was made. Then come the dimensions of Erin’s screen (834 by 1112). These are strange numbers, but I have found some players to show different screen sizes. I suspect that this comes from some smartphones with odd operating systems.
After three dashes comes the crucial information: encounter number and option. Erin played encounter number 297 and chose Option number 1. Lastly come some internal state values.
Data Analysis
I continue to gather data, but this analysis is based on the data obtained from December 11th through December 16th. During that time the program generated 2824 lines of data. Some of this data appears to have come from spiders who triggered only one line of data. There were 31 players who played through at least 3 encounters. The histogram of their usage, though, is disappointing:
As you can see, the great majority of players dipped their toes into the water and fled. This surprised me. The introduction to LMD went to great lengths to discourage gamers, warning them that LMD is not fun and requires a lot of reading. Yet most players decided they wanted to see for themselves, so they tried it out and then walked away after less than 40 moves — indeed, lots of the red group left after only 5 moves. Only one person made it all the way to the end.
At first glance, this appears to be catastrophically bad. EVERYBODY hates Le Morte D’Arthur! Yet I am unfazed by these results. I’m certainly not going to change the subject from the grim reality of 5th century Britain to knights in shining armor, flammivomous dragons, and damsels in distress. I’m not going to toss out all those tedious dialogues about the purpose of life and the meaning of death. I’m not going to replace all that dreary text with lively animations of flashing swords and explosions. This is art, not entertainment. If people don’t like it, well, that’s their problem, not mine.
Moreover, I suspect that there’s a strong selection effect here that biases these results. What kind of people read Chris Crawford’s Facebook posts are his website? Mostly gamers with more thoughtful inclinations — but gamers who seek better games, and LMD is not a game. But there’s another audience out there: an audience that enjoys reading and appreciates serious examination of the human condition. It’s a huge audience, and I think that some of these people would love LMD. My problems is that nobody in that audience knows about LMD.
This might seem like a classic marketing problem. The catch is, you can’t market art. Ads on TV won’t sell LMD; neither will ad images of big-breasted women and big-bicepped men. Art like this just has to seep into the pores of society and spread by word of mouth. That’s a long, slow process; I doubt that LMD will be recognized during my lifetime. But the nice thing about the web is that my website can long, long outlive me.
I wasn’t kidding when, 30 years ago, I declared that I was heading out into new territory. I complained that the games industry had become too god-damned civilized for my taste. Well, I suppose I got what I wanted. And I did slay the dragon.
Other observations
A few other tidbits emerged from the data. There was one person, “Reese”, who seemed to be playing while doing something else. Five or ten minutes would elapse between moves. Perhaps Reese had an extremely slow Internet connection.
There were about half a dozen players who were clearly thinking about their moves; they needed one or two minutes for each move. On the other hand, there were five players who raced through, making each decision in a few seconds. I suppose these people ignored all that dull text and kept clicking buttons randomly, hoping they’d get to something worthwhile.
The one player who reached the end did so in nine separate sessions, playing a few dozen moves in each session, then leaving for a while before returning.
I compiled all this data by hand, but I think I’ll need to write a program to carry out the analysis. Important point: always analyze the data by hand before trying to write a program to carry out the analysis. You have to get a feel for the data before you do anything else.