Docs:negtrin path: Add new thoughts

docs/design/negtrin-path.md

@@ -82,13 +82,26 @@ The fundamental unsolved problem is the contemplative evaluation whereby the can
 
 How does evaluation occur? Who does the comparison? We previously thought that the canvas does the comparisons but today we noticed that Director may be able to do them. In a sense, it makes more sense for Director to be the comparator because:
 
+#### Director as comparator:
+
 * If canvas merely helps director to transform mentents (we've finally found a good name), then director has more of an understanding of what specific instructions it's supposed to be giving to canvas.
   * Our previous model of having canvas hold both goals and mentents made it difficult to really understand how director could ever "know" what instructions to send to canvas.
   * With director holding goals, it can choose the types of scenes that need to be rendered in order to make comparisons and causal evaluations.
 * This makes the canvas implexing make more sense: the comparison is being done within director, but the canvas is making the comparison possible by rendering the mentents in orientations that enable canvas to compare them.
 * In a sense, this also makes it make more sense how compartmentalization occurs. A lot of the compartmentalization isn't within canvas, but rather it's between the director and canvas.
 
-### Where are contemplative qualia experienced?
+#### Canvas as comparator:
+
+In a sense, we can think of it as director partitioning the canvas, and finding some way to refer to the mentents within different partitions. Director can then implex the rendered output from canvas and compare that way. Perhaps the director only compares, and the canvas holds both the mentents and the goals that the mentents are being compared against.
+
+But how can the director know whether the goal currently loaded into the "goal partition" of the canvas is indeed the goal that it desired to compare against? The canvas is a distinct mentent space from the director's comparison stage. The director-comparator model has the advantage of ensuring that the director can be certain that its current goal matches whatever is being rendered by the canvas for it to implex from.
+
+What we can do is synthesize a bit:
+* Have a special shared region between the canvas and director called the "goalspace". This is where the director stores its current goal mentents and the canvas can read from this space directly. Some synchronization will be necessary between the two, but overall, it should be fine. Canvas will only ever need to read from it so we can prolly synchronize it with an RW spinlock, avoiding the overhead of sleeplocks (mutex, sem, etc).
+
+I think we settle on the director as comparator, canvas as orienter model, with a shared goalspace between them.
+
+### Where are introspective/contemplative qualia experienced?
 
 In introspection/contemplation, canvas renders the scene into the buffers. Director implexes from the buffers. So in that sense, contemplative qualia are experienced when director implexes them from the buffers.
 
@@ -96,3 +109,8 @@ For intrins, canvas renders them into the sense buffers and then director implex
 
 ### How does director's comparison yield the notion of intrinsic [un]desirability?
 
+Imagine director asking canvas to simulate the effects of some interaction that would bring about direct pleasure.
+
+Could it simply be that the director holds the goal of postrins as an innate comparison goal, and so when a comparison yields an intrin, it ends the need to compare, and the director just commits to action from then on?
+* Prolly not because we don't only figure out that an interaction will yield an intrin. With postrins for example, we figure out that they'll yield a postrin and then we spend a long time elaborately contemplating that postrin in self-indulgent scenarios.
+