Lilwik wrote:Kreistor wrote:And that's why I showed that Rob is not like them. He promotes speculation.

I'd say that writing so that your stories can be predicted by Occam's razor is worst thing you can do if you want to promote speculation. Occam's razor is not the only way to speculate; it's just the least interesting way. There's no challenge; you just count assumptions. In a story that uses Occam's razor as its standard of truth, as soon as Occam's razor has been brought to bear speculation becomes pointless.

I'm pretty certain Rob doesn't invent a bunch of speculation and then choose from them when he determines his story. That's what you just claimed he does. that's the only way to use the Razor during plot invention, to intentionally select places where low assumption speculations are avoided. I can't imagine the difficulty in performing that task reliably, but it seems an utter waste of time and energy to predict what the readers will choose to predict.

Lilwik wrote:Kreistor wrote:And many authors that ignore fan speculation are similarly random. They don't pre-select the dice against the speculators, because they ignore who is choosing what and which boxes are being chosen.

That's not random. They base the secrets of their stories on the underlying structure of the world they are writing about by means of careful thought and reasoning. That's not pre-selecting against most speculators, but it does tend to make things hard for speculators who insist upon counting assumptions instead of thinking about the world of the story.

Oh, goodness. You haven't detected Rob's pattern. Sigh. Here's what he does.

Rob needs Parson to appear to be a great general. There is an old RPG admonishment: do not play a character smarter than yourself. Robadmits to having very little experience with strategy games, and had none in tabletop games when he started this, so cannot appease the fans that expect Parson to be what Rob claims, because he isn't a great general himself. So, instead, he uses tricks to make Parson appear like a great general. One trick is to just use a huge amount of time to consider as many solutions as you can, but great "anythings" have inspiration that others never receive, so that can only go so far.

His most popular trick is to hide the rules from us. This achieves a second goal, BTW. In order to leave future storylines as flexible as possible, define the minimum you require to write the current story. So, he doesn't define caster Disciplines until he needs a caster to do something, then he chooses from them. Many will have some basic def'ns, but anything he doesn't tell us about can change, and we'll never know. By hiding the Rules, those of us with more strategy game experience can't introduce competing "brilliant" solutions and out-general Parson.

So, if you want to know what is going to happen, look at the rules he introduces. For instanfce, the page where he describes Dirtamancers can put out infernos. The moment you see that rule, you know that there will be an inferno and Sizemore will need to put it out. You might call that speculation. I call it meta-thinking a comic that by necessity must be minimalist in information disclosure. Consequently, my selection of the Throne move was grounded in a recognition of Rob's technique, and so I needed to have a theory where Sizemore could get to Spacerock. A new Side by Parson left Sizemore in a situation where he would have to Turn, which seemed unlikely at best, since Parson would have no way to contact him from Spacerock, without risking his head lopped off sticking it through the portal. Sizemore Turning to a Parson Side is not a 0% probability, but the roadblocks in the way of it happening seriously reduced it, to me. But not to others, so that faction persisted.

Another example is from the people that started reading the comic when it began. Several predicted that the volcano would explode the moment it was shown. It's meta-thinking, again, and it was correct. They used experience from other authors to raise the likelihood a tragic event would happen, because a place where it could happen was described. It's pretty close to an absolute: authors introduce very few unnecessary constructs, because they slow down the book and cost ink to print, and time to edit.

Lilwik wrote:Kreistor wrote:I gave you 7 dice types. The odds of any particular die being a D100 is, then 1/7, unless I also provided a distribution, which I did not.

In other words, you are assuming that if no distribution is given then none exists.

No. It is

not an assumption. It is Statistical theory, tried and true, pure mathematics. Used by everyone that uses Statistics for a living.

You are out of your depth. You left the world of Opinion, and entered the mathematical realm of Statistics where opinion is irrelevant. When you did, you entered a field in which I have a certain expertise and experience of use.

First, since I am trained, you reveal your lack of understanding of Statistical Methodology in every sentence. Second, you cannot convince me I am wrong, because I am trained by experts that acknowledged that I learned the subject matter, and you have no accreditation to counter that. Third, because this is a realm of study, you risk having textbooks dropped on your head.

You cannot win an argument in a mathematical discilpine without training in it. This is a branch of Mathematics, not paint on canvas where whatever you feel is true for you, or just because you make a fancy argument, you unravel a discipline in which geniuses disagree with you. There are methods and processes, proven and accepted, and these are pretty basic and hundreds fo years old in some cases. I do not have to explain or defend them, for them to remain true. And they are still true even if you decline to accept they are.

Lilwik wrote:Kreistor wrote:In the absence of a distribution, there are an infinite set of possible distributions. When you have a set to work with, you use the mean average of all possible distributions, which is trivial to calculate in this case. It comes out to the inverse of the number of possible selections, or 1/7.

So in other words, no distribution is the same as an even distribution. This all boils down to you taking all assumptions as equal.

False. Statistics tells us that in the absence of information to change the probability of any distribution being selected, treat all cases as equally likely. There is only one true distribution, not a probability of a distribution. In order to determine which distribution to use for our calculations, we turn to Statistical theory which is not incapable of giving us an answer. We are not predicting a chance of a predicted distribution being correct, but just determining which distribution to use, when facing an unknown distribution.

BTW, I am teaching: this isn't a debate anymore. I treat your statements as questions, not as a discussion of equals in which you might convince me I am wrong. I don't know how you forgot that I am trained in Statistics, but you have, and you don't get to tell me I'm wrong when experts have validated my capabilities. You may question my techniques, and in the end choose not to understand them, but if you do so you are choosing to not understand the mathematical discipline of Statistics, which is not based on Kreistor's opinion, but centuries of mathematical proofs, some of which are far too complex for this venue. Though not in this case: this stuff is basic. As obvious as it gets, really, when it comes to Stats.

By that thinking, a theory with two assumptions are always better than a theory with three assumptions, no matter how weak or solid those assumptions are. That's Occam's razor, but that's not real life, because in real life not all assumptions are equal.

In order for Occam's Razor to break down, the odds of something being false must drop drastically for the low assumption count cases. But since we determine those odds, we know it's a low chance in the first place. For instance, we can explain Charlie's knowledge of the MK by claiming "Charlie is really Sylvia." It's only one assumption, so we select it, right? No, we don't, because Wanda has control of Sylvia, at least as far as we can tell, and she knows when someone is not controlled by her, because of Awesomer's Turn. We can never make that chance be 0% until more information is revealed, but it's so low, we call it "wishful thinking". We reject obvious foolishness from Occam's Razor, even if it can't be proven to be false, because the probability is dismal.

Extreme cases are not evidence that you abandon a technique that works in non-extreme cases, when you are facing a non-extreme case. You only acknowledge that the technique has limits and stay within those limits. In engineering, we do not always use the precise equations that we know to be true of whatever we are modelling. Sometimes,the equations are so intensely complex to not be humanly implementable, so we have to spend long hours modelling them on computers. We may want to get done in a much shorter time, or we can't afford the modelling programs. So, instead, we develop approximate equations for regions of the curve that closely model that region (Drachefly's curve-fit), but are simpler for a human to calculate. For instance, an exponential can be approximated with two straight lines, and the only significant error will be found near the knee point, where if you need more accuracy, you can develop a series of smaller regions and lines. Even though those lines will be completely wrong for vast regions of the curve, they only need to be correct for the region of interest. So, extreme error in a region of unimportance is not evidence against the use of the approximation where it is accurate: just don't use it where it's not accurate. Engineers cannot achieve perfection, because nothing is perfect, so we deal in tolerances to error, which allows the use of these techniques to save time, money, and possibly even save lives.

In the "real" world, there is no probability of an assumption being true. It is true, or it is not, and we merely do not know which state it has. We can attempt to determine the probability of something being true, but there isn't some real probability in the world to calculate. We make informed estimates, which are no more "real" than Charlie. What is "real" is that the assumption is either true or false, and we hope someday to know which. Rob knows exactly how Charlie got his info, and presumably we'll find out one day, so there is a truth there to speculate about, not a probability of truth. The real erfworld knows how he go t the info: the probability of any speculation using an assumption is only our best guess at whether it might be true or false. We can assign probabilities to each assumption in any of the proposed answers, but there is a correct answer, with no probability of it not being correct. IN other words, one Speculation is in reality 100% correct, no matter what probability we assigned to it, or how many assumptions are involved. Occam's gets us past the conflict of not knowing which it is, by giving us a guide towards which choosing one, in order to continue forward in the face of the unknown.

Lilwik wrote:Kreistor wrote:Statistics is not paralyzed by an inability to know probabilities. We have solutions for that.

But the less you know about the probabilities, the less reliable the answers become, and Occam's razor is a technique that deliberately ignores all probabilities, even known probabilities that other people would use to make more reliable speculation.

No, Occam's is a technique you use when you don't know enough about the probabilities. If you know the probabilities, you can calculate accurate odds of all proposals, and choose the highest probability. You can create real chances of success in those instances, so you turn to statistical theory to guide your choice. You don't use Occam's to play Blackjack, you use probability, and developed Card Counting to affect those probabilities in the face certain non-random elements in the game. Occam's was invented to deal with discussions about demons and angels, which could not be tested for chance of existence, or their presence proven where they were being used to explain phenomenon. Odds were impossible, because evidence literally did not exist. And it is useful for Speculation, because there is equally no actual way to determine the odds of any one assumption being what is in Rob's head. Since the odds cannot be calculated, the Statistical method resolves to Occam's Razor, because the only thing we can evaluate is how many chances we are taking in each proposal, when we don't know what those individual chances are. You're proposing that when you knhow those chances, Occam's doesn't work, and you're absolutely correct. But in this comic, we don't know those chances, so Occam's is appropriate to apply.

Your opinion may guide you to conclude one Speculation is grossly unlikely, because one assumption is absurd. It's perfectly acceptable to exclude that proposal from your use of Occam's, but you need to be prepared to provide evidence of your estimation to those that contradict your estimate. Some assumptions seem absurd, until someone points out that you overlooked a section of the comic that provides it strong support, and that happens all the time here. (I got hit with the Turnamancy vehicle thing the other day. I'm vulnerable because I do not have an eidetic memory. Haven't seen many around here claim such. Anyone with one could cite page and write the text to support their positions without hunting the comic like I do, so it would be obvious that they have the capability.) So assigning a low likelihood requires more than just opinion, and the hive mind here helps us avoid that, by remembering parts of the comic we forget.

So, assign probabilities to assumptions if you like, but you'll need to be prepared to defend those estimates. I did so when I defended the Capital shift against the Parson's Side faction, using the same description of Rob's methods above, which skewed my estimate of the likelihood of the assumptions with related Rule references towards Capital Shift. People don't want to hear it, of course, but you'll need some rationality to an estimate, if you play the game.