Wednesday, November 12, 2014
And I'm awarding myself a paltry 50
LTBucks for the day!! Woohoo!
No, not terribly impressive, but in my still-high-on-cold-meds state, I at least managed to delve into some warp lane visual and audio candy. It's not awesome
yet, hence the minor award. But it's getting there! I also started on AI utilization of the rails. Again, not finished, but once healthy/lucid Josh returns, he should have no problem wrapping it up.
Instead of dwelling on that work, I'd like to speak a bit about my continued experiences with entering a half-dream state in search of inspiration, which is a fantastic activity when sneezing and sniffling cause one to just want to lay back in the chair and let the eyes droop shut. I'm not sure if it's due to the cold meds, or the continual practice of this strange art (as I said a while back, I've been practicing for about a month), or some combination of the two, but today brought a fascinating and highly-successful exploration of my subconscious -- by far the most successful I've had to date. Within it I found an abundance of inspiration and beauty related to LT, and even found some great technical insight. I realized that I need to do a better job of keeping in touch with this subconscious 'dream' of Limit Theory; after all, it's the dream that has pulled me so far in making this thing a reality. Interestingly, it was the first day wherein I had the idea to keep my keyboard on my lap as I drifted into dream land, which allowed me to record flashes of subconscious inspiration. Perhaps surprisingly, typing is physically-minimal enough that it doesn't disrupt a half-dream too much.
So you all think I'm crazy now, yes? ("Attention crew: this is Admiral Adama. DRADIS shows incoming wave of abstractness. Brace for impact!")
But seriously, I had some important revelations that started to tie together a great deal of my previous thinking on the relationship between objects and object types, attributes, and research (since research is intimately related to those attributes). I started to develop a mechanism for how all types
of things could be represented by attractors in a unified attribute space. For example, if we could define a set of unified attributes that can fully describe all weapons in the LT universe (a difficult problem, no doubt, because we would like to think that beam weapons are quite fundamentally different from missiles...but...are they really? Could we not quantify that difference using a unified set of attributes?), then we could encode the idea of weapon types
(pulse, rail, beam, etc.) by simply placing some number of 'attractors' in that space of attributes. Then, when we generate a weapon, we simply sample a random point in that attribute space, but apply the attractors afterward. The result is that the weapon gets 'pulled' towards what we might think of as the standard set of weapon types. This produces far more coherence than pure random generation. Yet, we can still maintain a unified continuity between all weapons, meaning that the implementation is shared, but also meaning that we can occassionally get some bizarre, exotic results (what would a weapon whose attributes are stuck halfway between a beam and a missile look like!?) I'm going to apply some more thought in this direction after I push this update out, because I know that my subconscious is onto something! This is something that I've hit on before with the research mechanism (research operates on attributes), but I've never seen it so clearly, nor have I ever realized that coherence in attribute space can be created by warping that space with attractors.
Although I probably won't be using this mechanism for weapons since I've already got so much code in place, I can tell you that I almost certainly will
be applying this technique to colonies! I've always been bothered by having to define some preset number of colony types. I would do much better, I think, to define a unified colony attribute space, and then place some attractor points within that space. This would give colonies a very real uniqueness. Despite uniqueness, we could still classify
them according to the closest attractor
! In other words, we can still say 'this colony is an agricultural colony,' because the attributes lie closest to the agricultural attractor point. Yet, the colony would remain unique from other agricultural colonies. You see, it's the difference between colony type as a set of values
and colony type as a classification based on a set of values
. Sure, it sounds abstract, but I promise, this is going to yield some very concrete benefits!! It's strange how so many patterns are starting to emerge in different places -- these attractors distort attribute space in the same way that PlateMesh warps distort geometry; the classification works in the same way that the game determines which zone you're currently in, etc. I guess it makes sense, though: at the end of the day, it's all just math!
Ah! Also forgot to mention that I started on 'custom items' today! We've got LTSL widgets, we've got LTSL objects, we've got LTSL AI tasks...so what's missing? LTSL item types! Time to complete the circle of rapid prototyping
Will speak more about this after update 21, at which time I'll begin to leverage it as I've been leveraging custom objects / widgets this month.
Alright, now that I've firmly established that I've gone off my rocker, I think I'll wrap this one up
Speaking of wrapping it up, as soon as my voice is back to sounding like a living, healthy human being, I'll be filming the update. I've pretty much got it all where I want it, and am ready and excited to let you all have a look at the beginning / continuation of the LT content explosion
I'm only going to put one reward on the table for tomorrow, because it's the only one that I really really
want at this point: 12PM EST
. Come on now. Tired of this devlog deficit! 100 LTBucks
, as usual. Go get 'em, monkey!
PS ~ Seriously, thanks for bearing with me through this log. I know that was a bit...crazy. But it's been a while since I've gone off on one of my crazy, abstract tangents, has it not?
Hopefully this rhinovirus isn't eating away at my rationality...
PPS ~ It occurs to me now that I should start calling this 'meditation' -- sound a bit less crazy, yes?