I'd like to make an analogy now about language, by thinking about language like a picture. And, supposedly, a picture is worth a thousand words, as the saying goes.

http://i.imgur.com/50jRyQ7.png
But not that picture. That picture's worth an incomplete sentence followed by a pregnant pause. Can you even tell what it is? You might be able to guess that at least part of it is a bike, based on the profile, and the fact that you can still kind of make out a wheel in the front....

http://i.imgur.com/F4mq9f5.jpg
...but any assumptions made about the rider(s) would have been way off.

Details are important. Especially if the subject matter is specifically regarding the granular details of an issue. Reducing the resolution of the question makes it impossible to ask a question about a "100x magnification" detail.

I have attempted discourse it the way you're suggesting before. But it's gotten to the point that, in frustration, I can anticipate how a conversation will go, the formula it will follow, when I begin with "lower resolution prose". Keeping with the metaphor, what ends up happening is, I ask about a detail in grid F,9 of the picture, and they say "Well that's a brown pixel." I then have to follow up with an other, slightly higher resolution, slightly less pixelated image, so as to not overwhelm them with the max resolution. They'll say "It's still a brown pixel." If I press, they might say, "Well.. actually I guess it's three brown pixels and one off-yellow pixel. But if you average the cube, it's still a brown pixel."

And we repeat that, me sending them a slightly higher resolution image, revising the question leading them into more detail.

After I've made them download about 5 progressively more detailed copies of the same file, they don't want the 6th and tell me to fuck off.

If I'm going to get fucked-off anyway, I'd rather it happen up front, so just give them the full resolution image right away, rather than wasting both our time trying to "dial in" the exact level of resolution they need in order to observe the detail in question.

I think that it can't be argued that the utility function of language is that of a communication protocol, in the truest sense of that term, from an information technology interpretation. I have an idea of X resolution in my mind, and I mean to upload it to someone else's mind. Due to the ponderously low resolution of the human larynx, there simply isn't the bandwidth or vocal range to perform that using a lossless protocol. But without Direct Neural Interface, that's what we're left with. Now, think about that undertaking procedurally. First the idea must be, through a conscious labor of thought, converted out of the animated, multidimensional, self-referential thoughtspace of abstraction, and shunted into a static, flat, language-compatible facsimile.

Already, from a fractal perspective, potentially infinite resolution has been lost. And we haven't even spoken yet. Next, the idea must be deconstructed in layers for broadcast. Should they be transmitted top down or bottom up? What are the dependencies? Does client and server in this data exchange share an experiential overlap? If not, how many conceptual dependencies must be included? How many second, and third tier dependencies? You're essentially required to decompile "idea source code" into a makefile, which will be broadcasted to an other thoughtware environment with unknown variables. Capabilities, experience, working world model, etc, could all be -- and most likely are -- completely divergent. Into this alien thoughtware environment, we send the idea seed that will be rendered by the recipient's procedural engineering incubator, to be reconstructed into, hopefully, some approximation of the original idea that we had in our head.

Without error correction.

There's no way to hold both ideas, the original and the copy, on the same thoughtware stage for comparison. Even from which axis in thoughtspace you began construction has an effect on the outcome, in the same that that filling a vessel with variable sized ball bearings one layer deep, packed to the walls, will create valleys between their curved edges, and necessarily effect the surface shape the next layer of ball bearings poured on top as they roll to equilibrium and fill the void spaces. With such an unacceptable signal to noise ratio, between the medium of communication and the unknown synchronicity between the encoding and rendering codecs at each end, the idea that one should choose "packets" with as high resolution and accuracy as possible should be taken as axiomatic. The notion that any meaningful communication can be expressed through the alternative is quixotic, at best.

...I have found that more... mundane? words tend to be afflicted with the lexicographical ailment of polysemy, which is the glutenous tendency some words have of taking an excessive number of definitions for themselves. Because of this, they lack granular detail, and are intrinsically noisy from a signal processing perspective. Too much chance of the incorrect definition being assumed at the recipient end, not to mention the tendency for ideologically biased recipients to capitalize on polysemous words by intentionally selecting to interpret them with the most inflammatory definition possible to further a personal or political agenda. This is why I select my words with unremitting and inveterate precision. There's no guarantee of reconstructing the source idea in the target mind, but the process needs all the accuracy it can get to overcome such a perilous signal to noise ratio.