Droid Engineer Archive

The Seven DEadly Sins of Droid Engineering: Sin
#2, Droid Experimentation

Disclaimer: I didn't gank this. And this is a long article. And
you don't have to read the justifications, but I recommend reading
the conclusion below.

Part
I: Technique [read more...]

Part
II: Are novices irrelevant? [read more...]

Part
III: Conclusion

CONCLUSION: It's my conclusion that the best
way to make a profession vibrant is to make the mastering of it
worthwhile. For DE, getting 5 more experimentation points isn't
enough, and it's not even strictly necessary. The number of
experimentation points matters not, only that they have a
significant effect. While experimentation “works”, the
effect is too small to make a difference. A probot with 825 HAM does
not have a significantly different effect on a battle situation than
a probot with 819 HAM. And that's the widest range I can achieve on
the probot. I've used them at both ends of that limited scale. Six
extra points in an 800+ HAM fighter is, at best, a rounding error in
your favor. If I had 10 points to experiment with, I might be able to
achieve a range of plus or minus 8-10 HAM points, which is a drop in
the bucket. Just getting those five extra points does not address the
heart of the experimentation problem.

I can make +3 and +4 crafting station modules, which
matters little next to 41.999 private crafting stations. It matters
little because I'm not certain those are even the same scales.
Apples and melons. But it doesn't matter to the user. That is the
heart of the experimentation problem. My customers prefer the +3 to
the -7 crafting station, and they think that maybe it does make a
difference but they aren't certain. It doesn't really matter to the
user. And crafting stations are where DE experimentation works best.
They serve their purpose, which is pretty cool, but they are not, at
the moment, ideal.

So here's a summation of today's Sin.

The second most deadly sin is to implement
experimentation that does not make a difference in user choices.

Product changes by virtue of experimentation must be
significant, both in the kinds of qualities that change, and range of
values that is achievable. The droid engineer may see that the final
product is slightly different, but if the user does not feel that
difference, then experimentation is either badly designed or badly
tuned.

What specific changes should happen?

First, when it comes designing components to be
experimented on, there are some better design choices and some worse
ones. A good design choice is to have three or four qualities, all of
which matter approximately equally, to experiment on. A great
design choice (if I do say so myself) would be to be able to optimize
specifically for, say, droid rolling/walking/flying speed, or droid
battery use efficiency, but to not be able to quite get 100% of both,
even for a master on the luckiest day with the best materials.
The user has to decide whether they want a droid that can keep up
with a burst run but that hogs batteries, or a droid that is energy
efficient yet slow. The fact that you can't have both is what makes
that choice meaningful.

Call me crazy, but these nebulous “quality”
and “durability” things just don't cut it. Durability
doesn't matter, and for half the droid components or the final
droids, you can experiment on qualities, get successes, but they
don't seem to correspond to a number. The component is what it is,
maybe it's a higher quality, but you have no way of examining or
monitoring it. That's just bad design.

If experimentation is not relevant for a droid or a
component, remove it. For the rest, make the effort to make
experimentation matter.

That's it for today. Tune in tomorrow to the next
article, wherein I discuss the Le Mort Du Droid! Bring the kiddies
and make a picnic of it!

I am really enjoying these articles Keep it up, neshura!

Not only do they address key issues we have as Droid Engineers, they make compelling arguments for and against their current state. I am going to say it again, if only the Devs / TH would read these, they would have a much better understanding of "where we are".

In defense of experimentation, once you have a decent amount of points (read are a master DE) you can improve the ratings of your modules (i.e. make you storage/data modules hold more, medical module have a higher rating). Granted once you are using level 6 modules you are already maxxed out on the ratings anyway, but for those who like to stack you can tweak the lower modules to get a max rating easier/cheaper.

The only problem is I'd rather have the points before they were useful than have them useful before I had them.

If experimentation mattered before we actually had the points to do it with that would be much much worse. I see this as the precursor to making it matter.

I don't see the issue on this one.