Comment by pg

As another poster alluded to, a practitioner of NLP would call this a "pattern interrupt".

  1) Build up an argument almost to a crescendo  
  2) Pattern interrupt ("it turns out")  
  3) New argument

A practitioner of NLP would say that this weakens the original argument because it puts an interruption between the stimulus and the response.

There is a shock to the system followed by the new argument that you would like the listener to adhere to. The shock makes the listener more suggestible.

That's the theory, anyway. The debate however is very controversial.

It tu... I mean, I suspect that it's ability to 'deceive' correlates strongly with how much you trust the author on that subject in the first place.

If someone tell me "turns out, the astrologers have been right all along!" ... that's not going to sway me unless they're sitting in front of a library of relevant peer reviewed evidence, and can guess which irrational number I'm thinking of.

I think I get it. In lieu of thinking, people use qualifiers as metadata for their belief filter that says how much belief they should assign to the proposition.

So if you qualify a proposition with "it turns out", that tells the filter that you should probably believe it. Maybe it triggers an assumption like "there is research backing this up."

Notice it sounds funny to say "It turns out that you're a dumbass," as the filter also flags boorish stuff and so it gets mixed signals there.

And it seems deceptive when you state contentious things this way, it's like you're trying to slip stuff past the filter. Unqualified declarative sentences generally seem to trigger this, people always accuse me of "stating my opinions as if they were facts" === "Flag contentious stuff for me so I don't have to think."

"... I'd be interested to see a list of all the 46 passages though ..."

There is more than 46 (725) ~ http://www.google.com/search?q=site%3Ahttp%3A%2F%2Fpaulgraha... If there is a pattern, could it be emerging from "asking questions", thinking, then finding non-obvious or surprising results? Reading through the authors site I found this gem also another explanation.

"... If one is convinced that mind is a computer program, or at least if one thinks of that as a fruitful metaphor, it may be helpful to think about these “thought-generating phrases” as function calls. They are like little labels that execute a useful module, or packaged set of computational instructions. And the point is, just having a label makes it easier to find the code, and therefore more likely that you’ll execute it. Which is exactly what you want if the code is worth calling. ..." ~ http://jsomers.net/blog/generating-thoughts

It quite possible the phrase, "it turns out" is also a thought generating phrase.

I agree, and so I collected most of them. I had a count of 36 just using the links on your essay page.

For any interested, I just submitted it to HN here: http://news.ycombinator.com/item?id=1163918

"I'd be interested to see a list of all the 46 passages though." -- the author links to http://www.google.com/search?q=site:paulgraham.com+%22turns+... which should allow you to see the words in a limited context

I think he wasn't calling the phrase itself deceptive. I think he's saying that you use it in an undisciplined way. From the final paragraph:

[the phrase is] useful in circumstances where you don’t have any substantive path from X to Y.

Nothing personal, but I think he's onto something. Especially if you take a more generous outlook and expand it to "circumstances where you don't demonstrate any substantive path from X to Y." (Even if somebody fails to demonstrate a substantive path, the path might still exist.)

Re: patterns, there was a blogger who did some kind of screen-scraping, machine learning, NLP analysis of my blog and claimed their analysis proved I was really a bot. I would take that kind of thing with a grain of salt.