everyone loves scatterplots

0 views
Skip to first unread message

Sean T Barrett

unread,
Nov 17, 2000, 3:00:00 AM11/17/00
to
Well, not really, I don't think anybody commented about them
last year, but what the heck, I make them to amuse myself.

These are scatterplots interrelating my votes, playtime, and
the comp scores. If you look on, you will be looking at ONE
person's evaluations. Not some kind of scatterplot of lots
of people's votes. Ok? Ok.

Key (please forgiving my oversimplification to "game" here):

- I quit the game because I wasn't enjoying the idea of playing more
+ I quit the game, not enjoying it, but planning on playing more later
@ I quit the game after getting as far as I think is possible
# I quit the game because I was stuck
x I quit, enjoying it, planning to finish later (e.g. time pressure)
* I finished the game
= Two games of type '-'

How I scored each game versus how long I spent on it (and why
I stopped when I did):

vote
10
9 * x
8 * * *
7 x -# *
6 # *+x=- #x* - *
5 + -= @==--= - +
4 + @#- - -
3 --
2 - * -
1 @
| : : ; : : ; : : ; : : | : : ; : : ; : : ; : : |
0:00 0:15 0:30 0:45 1:00 1:15 1:30 1:45 2:00

Outliers:
'x' in 0:10 column: the final game I played in the last minutes of the
comp (leaving myself with nearly 20 games to play
on the last day was unwise; 5 or so went unplayed)
'-' scoring 7: I quit playing this game because I didn't feel like
playing from the walkthrough, but thought it was a good
game nonetheless

Now, there's a nice trend visible in that scatterplot, but that's not
surprising; if I didn't like a game, I wasn't going to play it as long.
So, let's instead compare an "objective" rating of the game--the score
it got in the comp, rounded to nearest half--to the time I spent.

score
8 *
7.5 x
7 -* * x **
6.5 # * - -
6 - -* +*
5.5 + x -- # #
5 -- -- -+
4.5 =- -
4 * -
3.5 #- -
3 - - -- -
2.5 * - @
2 @-
| : : ; : : ; : : ; : : | : : ; : : ; : : ; : : |
0:00 0:15 0:30 0:45 1:00 1:15 1:30 1:45 2:00

Games towards the top-left are games that were widely liked but I only
played briefly. This can be because I got stuck (the top-left-most '#'
is "Shade") or because I got a bad impression early on (the '+' in 5.5
is "Enlistment", for which I've already posted the guess-the-verb problem
that made me quit; the '-' on 7.0 is Nevermore) or because it was a short
game (just to the right of "Shade" is "Rameses").

The lack of games to the bottom right suggests that either this strategy
led to me successfully avoiding spending much time on "bad" games, or else
that all bad games were short.

Finally, a scatterplot of my votes versus the comp scores:

Key:
- I quit the game because I wasn't enjoying the idea of playing more
+ I quit the game, not enjoying it, but planning on playing more later
@ I quit the game after getting as far as I think is possible
# I quit the game because I was stuck
x I quit, enjoying it, planning to finish later (e.g. time pressure)
* I finished the game
= Two games of type '-'

vote
10
9 *x
8 * *
7 x # - - * *
6 - + -#- *-x- # -x*
5 --- - *- = --- - - + *
4 @ - - - -
3 - -
2 - -*
1 @
| | | | | | | |
1 2 3 4 5 6 7 8
01234567890123456789012345678901234567890123456789012345678901234567890
comp score

Note that it's pretty meaningless to directly compare votes to comp average
scores, since different people assign different meanings to the scores,
and since often votes will range from 1-10 while final scores range
from 2-8. So the above scatterplot makes it more obvious the association:
comparing scores directly would be a straight line from 1,1 to 10,10 on
the above chart. Alternately we could fit a straight line to it. However,
there seems to be a bit of a curve here; I lean towards giving higher
scores to the middle of the pack; this no doubt reflects my bias to
reserve low scores for broken games as opposed to games I don't like.

The fragmenting on the top right suggests that there is wide agreement
about what is bad, but not wide agreement about what is good. (Well,
"wide" agreement is misleading--it's really agreement between me and
the rest of the world.)

So instead, we look at outliers in the above, which are generally the
endmost values. Assuming the comp scores are the "truth", I will
call my score "underrating" if I voted lower, and "overrating" if I
voted higher. I'm going to interpret the fragmented mess at the top
right as a very fat angled line, about four characters tall, and
consider those at the very fringes of it and beyond as outliers.

In the '3' row, on the right:

I voted 3, underrating "Desert Heat", which reflects the fact that
as far as I played through it, it was no more IF than "Little Billy".
Apparently it must branch further along.

In the 4 row, from the right:

I voted 4, underrating "The Masque of the Last Faeries"; apparently
other people had less problems with what I saw as sloppy mechanics
of both writing and programming.

I voted 4, underrating "Enlisted", for reasons described previously.

I don't think the leftmost "4" vote is a case of me overrating
"Comp00ter Game", since a "4" is not a good game in my book.

On the 5 row, from the right

I voted 5, underrating "Rameses", which I felt lacked the "I" in "IF".

I voted 5, underrating "The Best Man", which I had felt was a bit
too arbitrary and clunky in places.

I voted 5, underrating "The Planet of the Infinite Minds", which I
have posted about elsewhere. It just felt too arbitrary.
(It was the quintessential "drunk walk"--poke at things until
you have success, entirely without motivation.)

On the 6 row, from the left:

I voted 6, overrating "Prodly the Puffin". I actually scored Prodly
a 5, but gave it an extra point for making me laugh out loud
at the "bug in the hints" bit.

I voted 6, possibly overrating "Castle Amnos", which I had given up
on pretty quickly out of annoyance, but with plans to return to it.

From the right:

I voted 6 underrating "Ad Verbum", which I liked but I
felt the execution didn't live up to the concept (many untrapped
commands in the n,e,s,w rooms, some missing synonyms, some
arbitrary puzzles).

I voted 6 underrating "My Angel". I thought novel mode was insanely
cool and this was a great first attempt at it, but it really
needs more thought and polish into making the output read a
little more coherently. (And authors wishing to try this in Inform
might really want my varying-strings extension, which makes it
easy to create output statements which change on successive prints,
plug plug plug.)

I voted 6 underrating "Nevermore". I was unhappy with the game design.

The next on the right was "Shade", which I got stuck on so early that
it seemed to be a boring game about a "Burning Man" clone.

On the 6 row, from the left:

I voted 7, overrating "A Crimson Spring", marked 'x' indicating I needed
more playtime. This was the last game I voted on, under severe time
pressure, and I probably did overrate it.

I voted 7, overrating "The Ends Mean Escape", marked '#' indicating I
was stuck. I was enjoying the experience of the first room and never
got past it. Possibly my vote would have gone down had I seen more?
Possibly I should have voted it lower for having had a broken hint
system

On the right:

I voted 7, underrating "Kaged". I mostly felt this was a run-of-the-mill
dystopia, marred by some non-obvious puzzles and the comparison of my
problems with the red-coloring puzzle to the same problems with the
lower-quality "Withdrawal Symptoms". Disappointed that the mystery
wasn't ever really cleared up as far as I could tell.

On the 8 row, on the left:

I voted 8, overrating "Dinner with Andre", which I thought was a
marvelously silly little dish. "Being Andrew Plotkin" was funny,
but "Dinner with Andre" was a game.

I'll be writing short-paragraph reviews, but if any of the authors of the
aforementioned outlying games would like longer comments, I'll write them
on request (on the assumption that a deviant opinion might be more valuable
than a mainstream one).

Other notes:

My top two games made the top three.
My top three games made the top eight.
My top four games made the top eighteen.

My bottom two games made the bottom two.
My bottom four games made the bottom four.
My bottom five games made the bottom twenty-six.

The comp's top three games made my top six.

It would be interesting to compare my votes to the overall scores
in one sense--to see to what extent my tastes are abnormal--but
as I noted above the scatterplot conveys the information more
accurately than a numerical comparison. For example, I could
try to list which the games for which my votes had the widest
numerical deviation. Yet, *anyone* who did this would tend
to find that games they voted 1 or 10 on are going to have a
wide deviation, since the averages *always* pull those in.

For example, amongst my widest variations were my top
two votes, BAP and Metaorphses, both of which I deviated by
~1.7 points; but those two are in the top three, so the point
deviation isn't that meaningful. So really I need to fit a
curve to the last scatterplot, and then measure distance from
that curve or something. But *that* is more work than I'm
willing to do.

SeanB

Adam Cadre

unread,
Nov 17, 2000, 3:00:00 AM11/17/00
to
Sean Barrett wrote:
> I voted 7, overrating "The Ends Mean Escape", marked '#' indicating I
> was stuck. I was enjoying the experience of the first room and never
> got past it. Possibly my vote would have gone down had I seen more?

That was certainly my experience. In the middle of playing the first
segment, I was thinking I'd end up giving the game an 8 or a 9; upon
reaching the second segment and disliking it, I settled on a 5; after
finding out the *solution* to the second segment (which I found
nonsensical) and continuing onward, my score dropped to a low 3.

-----
Adam Cadre, Sammamish, WA
web site: http://adamcadre.ac
novel: http://www.amazon.com/exec/obidos/ASIN/0060195584/adamcadreac

Reply all
Reply to author
Forward
0 new messages