[Previous] Getting Elliot’s Attention | Home | [Next] Tracking Discussions

Discussion Trees With Example

When you have a discussion, it’s important to understand what is a reply to what, and what didn’t receive a reply (especially direct questions that aren’t answered).

To track this, draw a tree diagram. Put the initial thing someone said on top, then connect replies below it. Then for the each reply, put replies to it below it and connect them. And so on. It looks like this (real discussion, then tree):

Use abbreviated versions of what was said. Treat this like an overview, outline or notes. Make it condensed so it’s easier to see the whole discussion at once. Notes (text that doesn’t represent what someone said) can be put in square brackets. The tree helps show the structure of the discussion while having only short notes about what was said.

If it gets too complicated, you can split it into multiple diagrams. Write “subtree [name]” as a reply, then make a second diagram with that name which represents that part of the tree. It’s just the same as if you had one giant diagram except you took a part of it and moved it to a separate piece of paper or computer document. You can make documents that zoom in on specific parts of the overall discussion tree. You can also make an extra-abbreviated summary tree which leaves a lot out, then make some more detailed trees for some parts.

You should do something to indicate who said what, e.g. put their initials or use different colors.

It’s good to mark what didn’t get a reply and non sequiturs (comments that aren’t responsive, don’t engage with what they reply to). You could also mark direct questions, or at least direction questions that weren’t answered.

In my example, a green outline is Jack Dorsey, red is me, and black is an anonymous poster named A. Bold indicates a direct question (I paraphrased some things as questions but only bolded if it was a question in the original text). Dotted lines are non sequiturs. Ovals are statements that were replied to and rectangles are statements that were not replied to.

You can keep a tree in chronological order if you extend the lines between replies. Each row can be a message someone sent. If someone replies to an old point, draw a long line from it down to the current row. You can draw horizontal lines the show the rows. This will help with complicated discussions. Look at how my example tree is organized in rows. You never see claims from the same person in the same row, and every row corresponds to a specific message (I wrote three messages in the discussion and I have three rows, same for A).

Trees help you understand the discussions you have. Practice making trees for many of your discussions until it’s easy. Also practice doing it with other people's discussions. (If other people's discussions are easier because you're less emotionally involved or biased, start there; if it's harder because you understand what's being said less, start with your own.) Mentally keeping track of trees like this is what people who are good at discussions do (except when they actually write notes). If you write them down a bunch of times, you’ll get way better at remembering them.

When you have a difficult discussion with someone, if you both share your tree diagrams, you can compare and see where you view the discussion differently. This helps clear up misunderstandings and other problems.

Tree Analysis

The tree diagram makes it easy to see that A wasn’t responding to most of what I said (look for the red rectangles and the dotted lines). You can also see the two things from A that I didn’t reply to. And you can see what happened with direct questions: first, no real answer, just a vaguely implied answer that doesn’t make sense (I asked the point of what he was saying and he implied no point) and then a non sequitur reply, that does not answer the question, to my followup question trying to ask the same thing again.

It’s hard to perfectly represent discussions as summary trees but you can represent a lot of information this way. It’s useful even if it’s not 100% complete. In this case, the tree leaves out an issue that helps explain why I didn’t reply to the claim that debates are irrational.

I said:

You haven't given reasons nor any way for me to learn that you're right and change my mind.

And A replied criticizing me for mentioning debate, saying:

learning from each other is what matters.

I had just complained about the lack of any opportunity to learn from him, and then he criticized me because, allegedly, I wanted to debate in a non-learning way. That’s unreasonable and it’s part of a pattern where he didn’t engage with any substantive thing I said (look at all the square rectangles, plus what happened with my direct questions).

Discussion trees are literally and technically equivalent to bullet point outlines with nesting (indenting). You nest/indent replies under what they reply to. That represents the identical information as a tree with lines indicating what is a reply to what. If you don’t understand this, practice creating both the tree and the outline until you do understand.

Making Trees

You can make tree diagrams with pencil and paper, art apps (FYI vector art apps like Affinity Designer make more sense than pixel or photo based apps like Photoshop, and more basic tools can work too, and there are mind mapping and diagramming apps), OmniGraffle, or Graphviz. For info on generating tree diagrams from s-expressions, see my email reply to Justin (who found a website which does it), sharing my Ruby script which converts s-expressions to Graphviz files. Here’s the s-expression I used to create the example tree:

("No political ads on Twitter"
    ("social status, favors, friends, pull"
        ("money shouldn't buy influence"
            "no info that could change my mind")
        (disagree
            "no reasons"
            ("debate?"
                "debates are irrational, aren't you a Popperian?")
            ("point?"
                ("[implied] there is no point"
                    ("purpose of contradicting me?"
                        "opinions are allowed here")))))
    "Less upward mobility"
    "Can't put money where mouth is"
    "Read Atlas Shrugged")

It’s worth learning to write trees as s-expressions. s-expressions are a general purpose intellectual tool. They’re a way of representing structured information/data.


Update:

See the Discussion Trees blog category for more tree examples.

See Mind Map software review for software choices.


Elliot Temple on October 31, 2019

Messages (44)

This dialogue tree thing encompasses beautifully how vastly incompetent you can be about social things.

Could be kind of useful for formal debates though. An interesting idea for organization but a horrible way to try to understand another human being.


A at 6:40 PM on October 31, 2019 | #14074 | reply | quote

It should read how vastly incompetent one* can be about social things. Communication is difficult.

I reiterate it seems like a good way to get organized, especially if you're thinking in formal debate scenarios.

However, to best understand someone else, this seems like a terrible idea.


A at 7:35 PM on October 31, 2019 | #14075 | reply | quote

Noted by the fact that you confused "I am not interested in debates" with "debates are irrational."

Words matter, you ought to be careful.


A at 7:45 PM on October 31, 2019 | #14076 | reply | quote

#14076 Please don't post misquotes here.

PS You seem hostile. You don't seem like you want to reach an understanding with me and you're just venting rudely.


curi at 7:56 PM on October 31, 2019 | #14077 | reply | quote

Bro, you literally misquoted me.

It is in the very post you wrote, I literally said I am not interested in debate.

and you posted

> "no reasons"

("debate?"

"debates are irrational, aren't you a Popperian?")

("point?"

You're confused.


A at 8:04 PM on October 31, 2019 | #14078 | reply | quote

how about you follow your own rules and don't misquote people.

You should use copy paste instead. That way you can't misquote someone

> also debates are not interesting to me

is not the same as stating that debates are irrational. You should not misquote people, it is unethical.


A at 8:06 PM on October 31, 2019 | #14079 | reply | quote

I think you should do a post mortem on the mistake you made here. How will you prevent yourself from misquoting other people?

How are you going to stop yourself from assuming hostility when people are rightly defending themselves against being misquoted?


A at 8:12 PM on October 31, 2019 | #14080 | reply | quote

I didn't misquote. Those aren't quotes. You're sure yelling a lot just because you don't know how s-expressions work and what strings are.


curi at 8:26 PM on October 31, 2019 | #14081 | reply | quote

Don't evade. You're being dishonest.

"not interesting to me" is not the same as saying that debate is irrational.

no matter how hard you try to spin it. You presented what you said accurately but somehow its okay to misquote what I said?

You should stop being dishonest. Should do a post mortem on how to prevent yourself from making these mistakes, and why you're doubling down on your mistakes.


A at 8:31 PM on October 31, 2019 | #14082 | reply | quote

> Bro, you literally misquoted me.

> It is in the very post you wrote, I literally said I am not interested in debate.

> and you posted

>> "no reasons"

> ("debate?"

> "debates are irrational, aren't you a Popperian?")

A, the words inside "" that you mention in post #14078, and confuse for quotes, are strings. See https://en.wikipedia.org/wiki/String_(computer_science)

I think it's clear that does are not quotes as it is under the title "Making Trees" (explaining how to make the diagram), is structured as something resembling code to most people that have seen code in some way before, and is following the sentence:

> Here’s the s-expression I used to create the example tree:


N at 12:13 AM on November 1, 2019 | #14083 | reply | quote

Forest for the trees my dude. You're completely missing the point that Temple is purposely misrepresenting what A said. This is dishonest.

This is something that curi commonly does. If he spent half the time and effort that he does in trying to prove himself right into understanding people who are disagreeing with him, he would actually learn.

If you can't state your opponent's argument in a way that they would endorse it, then you have not understood it. Temple fails to understand the people he disagrees with him most of the time.


Philosophy is meant to be fun. at 7:03 AM on November 1, 2019 | #14086 | reply | quote

#14086

> Forest for the trees my dude. You're completely missing the point that Temple is purposely misrepresenting what A said. This is dishonest.

Did the possibility occur to you that Elliot believes that it's not a misrepresentation?

> If you can't state your opponent's argument in a way that they would endorse it, then you have not understood it. Temple fails to understand the people he disagrees with him most of the time.

Note that Elliot's outline is for *himself*, not for anybody else. The outline is how *he* understands the discussion. It's *his* perspective.

If somebody wanted to understand elliot's view, they could ask to see elliot's outline (if he made one).

If somebody wanted to get on the same page with Elliot, they could share their own outline and ask to see elliot's outline (assuming both parties made one). and then each party could more easily identify disagreements between them.


Anon446 at 2:47 PM on November 1, 2019 | #14096 | reply | quote

Sharing a discussion tree is a good way to get more attention from me, if you want that but don't want to request a debate. It's one of the ways you could put more effort into a conversation to earn some effort/attention on my part.


curi at 10:59 AM on November 2, 2019 | #14104 | reply | quote

#14104 How much effort do you put in understanding people you disagree with. I've never seen you steelman someone you disagree with. That's a bad sign.


Anonymous at 11:57 AM on November 2, 2019 | #14105 | reply | quote

#14105 A huge amount. Do you have an example where you think I failed to do this but should have?


curi at 12:00 PM on November 2, 2019 | #14106 | reply | quote

curi at 12:03 PM on November 2, 2019 | #14107 | reply | quote

If you can't restate your opponent's argument in a way they would endorse it, then you have not understood it.

This is bad because if their argument is bad, you don't understand why or how it is bad. And if their argument contains truth, you would miss that truth.

That is all there is to steelmanning, restating the other's argument in the best light possible. Then criticizing that. Does the link you provided address what I said here?


Anonymous at 12:09 PM on November 2, 2019 | #14108 | reply | quote

1) Repeating: Do you have an example where you think I failed to do this [putting effort into understanding the other person's ideas] but should have?

2) Yes the link is still relevant.

You have not engaged with either thing I said.


curi at 12:14 PM on November 2, 2019 | #14109 | reply | quote

> You have not engaged with either thing I said.

Sadly, neither have you. Perhaps you could use this chance to learn how to steelman and attempt to restate my argument in a way that I would endorse it.


Anonymous at 12:17 PM on November 2, 2019 | #14110 | reply | quote

I did engage with you. You asked if I'd put effort into something. I told you I had and attempted to progress the discussion to find out what we disagree about (it looks like there is a disagreement). You brought up steelmanning. I linked you to my thoughts on that. That's engagement. You, on the other hand, have no specific criticisms, don't want to form any, and don't want to read what I have to say about your topic.

I think you're writing low quality comments, so I'm done unless you're willing to request a debate. https://elliottemple.com/debate-policy


curi at 12:21 PM on November 2, 2019 | #14111 | reply | quote

I do not endorse your restatement of my argument or what you think my position is.

I am not requesting a debate. I think a policy for me, since you have so many it is only fair I share one, is that the person I debate or argue with is capable of intellectual honesty. This includes not strawmanning my arguments and misrepresenting my views.

You have not shown to have the ability to argue in good faith and failed miserably at steelmanning my argument.

As restated here for your convenience:

>If you can't restate your opponent's argument in a way they would endorse it, >then you have not understood it.

>This is bad because if their argument is bad, you don't understand why or how it >is bad. And if their argument contains truth, you would miss that truth.


Anonymous at 12:26 PM on November 2, 2019 | #14112 | reply | quote

I agree with Anon.

Refuting straw mans does nothing for our own argument. This is why The principle of charity or Steelmanning is so important.

*The Principle of Charity is a methodological presumption made in seeking to understand a point of view whereby we seek to understand that view in its strongest, most persuasive form before subjecting the view to evaluation.

While suspending our own beliefs, we seek a sympathetic understanding of the new idea or ideas.

We assume for the moment the new ideas are true even though our initial reaction is to disagree; we seek to tolerate ambiguity for the larger aim of understanding ideas which might prove useful and helpful..

Emphasis is placed on seeking to understand rather than on seeking contradictions or difficulties.

We seek to understand the ideas in their most persuasive form and actively attempt to resolve contradictions. If more than one view is presented, we choose the one that appears the most cogent.*

Can you learn to do this curi?


Philosophy is meant to be fun. at 12:33 PM on November 2, 2019 | #14113 | reply | quote

#14113 You didn't engage with my FI post on this topic, nor did you criticize any example where you believe I did something wrong – even after I'd already just brought up those issues (and you agree with anon but gave no defense of anon doing those those things).

And you are strawmanning me by seeming to assume I disagree with or don't know stuff, much of which I know and agree with.

So see https://elliottemple.com/debate-policy


curi at 12:37 PM on November 2, 2019 | #14114 | reply | quote

#14113 Well said.


A at 12:54 PM on November 2, 2019 | #14115 | reply | quote

> And you are strawmanning me by seeming to assume I disagree with or don't know stuff, much of which I know and agree with.

I apologize not my intention. Please tell me what you agree with from what I said and we can build from there.

I do not have a blog so if that disqualifies me so be it.


Philosophy is meant to be fun at 4:57 PM on November 2, 2019 | #14122 | reply | quote

#14122 Do you want to debate this to a conclusion (agreement or explicitly stated impasse chain) and satisfy all the other conditions?


curi at 5:24 PM on November 2, 2019 | #14123 | reply | quote

I am not a very good debater but...

I think this could be an interesting line of discussion. However, I do not check back frequently and I may not respond again until a few days later.

If you're fine with that we can try.

Personally, I prefer email for serious discussions.


Philosophy is meant to be fun at 5:45 PM on November 2, 2019 | #14124 | reply | quote

#14124 Are you aware that I have an email-based discussion forum?


curi at 5:47 PM on November 2, 2019 | #14125 | reply | quote

I'll sign up for that instead. Seems like you have a ton of rules for that as well.

You have a lot of hoops that people have to go through to just have a discussion.

I sympathize. Most people suck so putting up all these gates is probably rational.

I'll be talk to you soon.


Philosophy is meant to be fun. at 5:56 PM on November 2, 2019 | #14126 | reply | quote

I'll be talking *


Anonymous at 5:57 PM on November 2, 2019 | #14127 | reply | quote

#14126

> I sympathize. Most people suck so putting up all these gates is probably rational.

OK I'm glad you understand. The gates on the email group are basically just formatting emails. The group would not really function without that. People who don't understand how nested quotes work in email can use the website.

It's not just that most people suck but also that I've already written ~50,000 discussion messages talking with most people. So they now suck *repetitively* which is a lot less interesting.

Responding every few days is fine.


curi at 6:19 PM on November 2, 2019 | #14128 | reply | quote

#14126 What I find (and have mentioned before and am currently writing a blog post about) is that other intellectuals have gates too. But they use *unwritten rules*. Their gates/rules are much harder to satisfy, and actually unreasonable, but they obscure this and prevent critical analysis by refusing to say how they work. I think my gates are actually very minimal by comparison, but they look like a lot because I openly stated them.

I really dislike unwritten rules in many contexts, e.g. a government that doesn't operate by clearly written laws or forum moderators who moderate posts according to unwritten rules. I think this is a major problem with approximately all public intellectuals in general.

With intellectuals, there are two separate sets of rules which are unwritten. The first is the rules for getting a discussion and the second is the rules for how rational discussion works, like what actions are appropriate or inappropriate during a discussion (including, as a key issue, how to rationally end a discussion without mutual agreement).

The lack of clearly documented rules allows for tons of bias. It doesn't just allow bad rules to be hidden, it also allows for rules to be applied inconsistently. It prevents accountability.

Popper, Rand, and other people *who I greatly respect* did not specify rules like this either. I think/hope they would have if I'd suggested it to them. I think it's an important insight about how to be a rational intellectual – it's actually a significant piece of epistemology. The internet has made it more important because it's greatly increased opportunities for discussion with the general public instead of just within a social circle or peer group.


curi at 6:31 PM on November 2, 2019 | #14129 | reply | quote

#14129 Not who you're talking to but I completely agree with everything you said here.

I don't think it changes the steelman part of the discussion but I agree with this response.

Do you have an example of you steelmanning an argument you disagree with? That would change my mind and I'd admit being mistaken.


A at 6:39 PM on November 2, 2019 | #14130 | reply | quote

#14130 Here's an example from Discord today. By "He", Mingmecha is referring to the author of #14105

Mingmecha: He is anti- individual anti- personal responsibility.

curi: what do you mean?

curi: which statement(s) are you getting that from?

Mingmecha:

> Sadly, neither have you. Perhaps you could use this chance to learn how to steelman and attempt to restate my argument in a way that I would endorse it.

Mingmecha: He wants you to do all the work he doesn't want to have to explain himself specifically he thinks it's a group effort

curi: oh do you mean he's bad at responsibility rather than anti (against it intellectually)?

Mingmecha: Yeah maybe I phrased it wrong

In this exchange, curi disagreed with what Mingmecha said, as written. After a little clarification, curi came up with a variant of Mingmecha's that he thought was better – a "steelmanned" version. Mingmecha accepted it.

You should try to answer curi's question to you by attempting to point out a case where you think he made a mistake. You have not made any substantive critical statement and have resisted doing so, in addition to resisting reading what curi had to say about steelmanning (how can you steelman his views if you don't even read what they are?).


Howard Roark at 6:50 PM on November 2, 2019 | #14132 | reply | quote

I think that is sort of what I mean. However, they are in general agreement. I mean when he talks to people he disagrees with. For example, I feel like he strawmanned me.

As did the person in that discord conversation.

I am the case you keep asking for. I'd like to see an example of curi steelmanning someone he has substantial disagreements with.


A at 7:01 PM on November 2, 2019 | #14134 | reply | quote

#14134 You are *moving the goalposts*. Also:

> For example, I feel like he strawmanned me.

Give a quote and say how that text strawmanned you or what's wrong with it and what you think he should have done.


Howard Roark at 7:11 PM on November 2, 2019 | #14135 | reply | quote

> Give a quote and say how that text strawmanned

Quote: Restate my argument so I know you understood

Discord user says: He's anti-individual.

lul


Anonymous at 7:49 PM on November 2, 2019 | #14136 | reply | quote

#14136 I would prefer if you don't post comments on my blog. Please leave.


curi at 7:53 PM on November 2, 2019 | #14137 | reply | quote

After you can explain how that's not a blatant strawman.


Anonymous at 7:57 PM on November 2, 2019 | #14138 | reply | quote

> I think that is sort of what I mean. However, they are in general agreement. I mean when he talks to people he disagrees with. **For example, I feel like he strawmanned me.**

This is vague. What did he strawman, according to you, A? What did you actually mean? I do not think that anyone actually understands what you are referring to by "I feel like he strawmanned me". At least I do not understand it. Would you care to explain?


N at 12:13 AM on November 4, 2019 | #14150 | reply | quote

After testing, I cannot recommend OmniGraffle unless you just want to do trees in that app and export as images or PDFs.

If you want to e.g. write an outline in a markdown file then get it imported, or export a tree back to markdown, then stay away.

It's like a rich text editor with all the typical problems.

OmniGraffle can't import text files but OmniOutliner can. But it can't import e.g. italics or bold from my Ulysses docs. And it can export plain text that works in Uylsses but with formatting lost.

I manually fixed some bold and italics in Outliner after importing. Then I imported into Graffle and it lost the bold and italics when importing from their own app into their own app. I also had to spend a while getting the imported doc to be usable at all, e.g. it defaulted to text overflowing out of the rectangles around it. And lots of minor hassles I didn't go into. Still better than most of the mindmapping and diagramming apps out there.


curi at 2:29 PM on November 10, 2019 | #14278 | reply | quote

GraphViz doesn't support automatic wordwrap. I forgot about that but it basically makes it unusable for me for large trees where I write a whole paragraph in some nodes.

I tried xmind (like omni you can import from a markdown list and lose formatting but keep your nesting structure) but don't like the spacing it generated and it doesn't have many features or much control over spacing.

I might end up putting up with Omnigraffle. Best option currently. I can import to Outliner from a markdown bullet point list outline, lose formatting, import straight to Graffle (lose formatting again), and then get an OK output without individually editing nodes and then export to svg. So I'll stick to almost entirely plain text with this approach.

I've done some trees in Affinity Designer and it's too much work laying everything out manually for a big tree with a lot of nodes. I think any kind of freeform art program isn't going to satisfy me for trees where I write a lot of text.

SimpleMind doesn't seem to even do trees, only maps with the root node in the middle and branches going to both sides. Xmind defaulted that way but let you change the structure.

Taking suggestions.


curi at 4:09 PM on November 10, 2019 | #14279 | reply | quote

curi at 1:32 PM on March 27, 2020 | #16164 | reply | quote

Want to discuss this? Join my forum.

(Due to multi-year, sustained harassment from David Deutsch and his fans, commenting here requires an account. Accounts are not publicly available. Discussion info.)