Self Standing Evidence Is Impossible

From SelfStandingEvidence:

SelfStandingEvidence is impossible. Every word in the dictionary is defined by other words in the dictionary. All theorems depend on axioms.

As stated below, nobody is suggesting it in absolute or extreme terms. There should be some "axioms" or definitions we can assume the reader knows without proving upon each usage that 2 + 2 = 4. But which ones has not been agreed on yet. See WikiPrerequisites. If you write "I fell off my chair", we assume that everyone knows what "fell" and "chair" means.

Right, since there is no such thing as SelfStandingEvidence let's not use the term. It has no meaning and adds no value.

Well, if it is a useless concept it is probably because nobody will agree on the prerequisites, not because such wouldn't be helpful.

No, it's a useless concept because it doesn't exist, regardless of any prerequisites. You just said so yourself.

Huh?

If evidence relies on collective agreement to some set of prerequisites (which you argue for above) then it doesn't stand by itself.

Like I keep saying, it is not an all-or-nothing thing. If you want a longer title that makes that clear, be my guest.

If "self-standing evidence" doesn't stand by itself, it isn't a thing at all. I don't want a longer title. I want to delete this page.

Evidence A can be MORE self-standing than evidence B.

How is that? If evidence A relies on 1 axiom and evidence B relies on 2 axioms, why is A any more "self-standing"? It still relies on an axiom.

Something that relies on fewer external axioms is generally "better". We are striving for better here. Perfect would be nice, but it ain't gonna happen.

Do you have any evidence that "something that relies on fewer external axioms is generally better"? And "better" in what sense?

A form of OccamsRazor.

Occam's Razor says that if you can eliminate an axiom then you should. It says nothing about fewer axioms making something "better".

Are you saying you have no preference for an idea that uses only 10 axioms over one that uses 1000?

Yes, that is what I am saying. There are very simple ideas that require little previous knowledge to understand and apply. There are very complicated ideas that require a great deal of previous knowledge to understand and apply. The simple ideas aren't "better" than the more complicated ideas. Often times they are less accurate and/or less useful.

All else being equal, would you take the 10 or 1000?

If the same proof could contain 10 or 1000 axioms, then I'd chose the proof that used 10. That's OccamsRazor. But we aren't talking about the same proof, we're talking about different arguments. I wouldn't reject one argument because it relied on more facts than another argument.

Okay, but what about argument A that has 50 axioms here and 50 external, versus one that has 90 here and 10 elsewhere. From a practical perspective, readers are going to pay more attention to the second. We should be nice to readers.

You lost me. Are you saying the people who read this wiki are likely to read only this wiki, and that they haven't studied computer science elsewhere?

I believe there should be WikiPrerequisites that expect a minimum knowledge. This wiki should target practicioners, not really newbies to computer science. There are better places for that.

And if you are saying that, are you further arguing that we should only discuss ideas that are already heavily documented here?

No, I am saying that one should not insult an opinion based only or mostly on external evidence. Thus, nobody should say, "If you knew what I know, you would know that you are flat wrong." Under SelfStandingEvidence thinking, that is not acceptable wiki edicate.

WikiPrerequisites are spelled out on that page (web browser, internet access, working knowledge of English). Will you answer my questions?

Which ones were not answered? Information that relies on too many external axioms is of less use to most readers. I am just the messenger of that fact of human nature. As an example, my "side flag" typing model described under AlternativeTypeDefinitions may (allegedly) be less thorough than Reynold's typing model, but more may find it useful because they don't have to deal with massive BookStops. They get a usable model (perhaps a UsefulLie like Newtonian physics) much quicker. Suppose they are getting a 95% useful model at a cost of 10 units of energy as opposed to Reynold's 99% useful model at the cost of 200 units of energy.

Three words: CogitoErgoSum

which is only a self standing argument in that it's circular. Have a look at

http://en.wikipedia.org/wiki/Talk:Cogito_ergo_sum unconvincing

CategoryDiscussion