How to come to better conclusions by playing steelman solitaire

Cross-posted from the Char­ity En­trepreneur­ship blog.

Ac­knowl­edge­ments. I’d like to thank Spencer Green­berg for both in­spiring the origi­nal idea with Clearer Think­ing’s Belief Challenger tool and for com­ing up with a much bet­ter name for the con­cept than my origi­nal “steel­man­ning back and forth”.

I have a tool for think­ing I call “steel­man soli­taire” that I have found comes to much bet­ter con­clu­sions than do­ing “free-style” think­ing, so I thought I should share it with more peo­ple. In sum­mary, it con­sists of ar­gu­ing with your­self in the pro­gram Work­flowy, al­ter­nat­ing be­tween writ­ing a steel­man of an ar­gu­ment, a steel­man of a counter-ar­gu­ment, a steel­man of a counter-counter-ar­gu­ment, etc. (I will ex­plain steel­man­ning later in the post; in brief, it is the op­po­site of a straw­man ar­gu­ment. Steel­man­ning means try­ing pre­sent the strongest pos­si­ble ver­sion of an op­pos­ing view.) In this blog post I’ll first ex­plain the broad steps, then list the benefits, and fi­nally, go into more depth on how to do it.


  1. Struc­ture forces you to do the thing you know you should do any­way. Most peo­ple read­ing this already know that it’s im­por­tant to con­sider the best ar­gu­ments on all sides in­stead of just con­sid­er­ing the weak­est on the other. Many already know that you can’t just con­sider a counter-ar­gu­ment then con­sider your­self done. How­ever, it’s easy to for­get to do that. The struc­ture of this method makes you much more likely to fol­low through with your ex­ist­ing ra­tio­nal as­pira­tions.

  2. Clar­ifies think­ing. I’m sure ev­ery­body has ex­pe­rienced a dis­cus­sion that’s gone all over the place, and by the end you’re more con­fused than when you started. Points get lost and for­got­ten while oth­ers dom­i­nate. This ap­proach helps to or­ga­nize and clar­ify your think­ing, re­veal­ing holes and strengths in differ­ent lines of thought.

  3. More likely to change your mind. As much as we as­pire not to, most peo­ple, even the most com­pe­tent ra­tio­nal­ists, will of­ten be­come en­trenched in a po­si­tion due to the na­ture of con­ver­sa­tions. In steel­man soli­taire, there’s no other per­son to lose face to or to hurt your feel­ings. This of­ten makes it more likely to change than your mind than a lot of other meth­ods.

  4. Makes you think much more deeply than usual. A com­mon fea­ture of peo­ple I would de­scribe as “deep thinkers” is that they’ve of­ten already thought of my counter-ar­gu­ment, and the counter-counter-counter-etc-ar­gu­ment. This method will make you re­ally dig deeply into an is­sue.

  5. Deal­ing with steel­men that are com­pel­ling to you. A prob­lem with a lot of de­bates is that what is con­vinc­ing to the other per­son isn’t con­vinc­ing to you, even though there are ac­tu­ally good ar­gu­ments out there. This method al­lows you to think of those rea­sons in­stead of get­ting caught up with what an­other per­son thinks should con­vince you.

  6. You can look back at why you came to the be­lief you have. Like most in­tel­lec­tu­ally-ori­ented peo­ple, I have a lot of opinions. Some­times so many that I for­get why I came to hold them in the first place (but I vaguely re­mem­ber that it was a good rea­son, I’m sure). Writ­ing things down can help you re­fer back to them later and re-eval­u­ate.

  7. Bet­ter at com­ing to the truth than most meth­ods. For the above rea­sons, I think that this method makes you more likely to come to ac­cu­rate be­liefs. ​


Straw­man­ning means pre­sent­ing the op­pos­ing view in the least char­i­ta­ble light, of­ten so un­char­i­ta­bly that it does not re­sem­ble the view that the other side ac­tu­ally holds. The term of steel­man­ning was in­vented as a counter to this; it means tak­ing the op­pos­ing view and try­ing to pre­sent it in its strongest form. This has some­times been crit­i­cized be­cause of­ten the al­ter­na­tive be­lief pro­posed by a steel­man also isn’t what the other peo­ple ac­tu­ally be­lieve. For ex­am­ple, there’s a steel­man ar­gu­ment that states that the rea­son or­ganic food is good is be­cause mo­nop­o­lies are gen­er­ally bad and Mon­santo hav­ing a monopoly on food could lead to dis­as­trous con­se­quences. This might in­deed be a be­lief held by some peo­ple who are pro-or­ganic, but a huge per­centage of peo­ple are just fal­ling prey to the nat­u­ral­is­tic fal­lacy.

Nonethe­less, while steel­man­ning may not be perfect for un­der­stand­ing peo­ple’s true rea­sons for be­liev­ing propo­si­tions, it is very good for com­ing to more ac­cu­rate be­liefs your­self. If the rea­son you be­lieve you don’t have to care about buy­ing or­ganic is be­cause you be­lieve that peo­ple only buy or­ganic be­cause of the nat­u­ral­is­tic fal­lacy, you might be miss­ing out on the fact that there’s a good rea­son for you to buy or­ganic be­cause you think mo­nop­o­lies on food are dan­ger­ous.

How­ever, and this is where steel­man­ning back and forth comes in, what if buy­ing or­ganic doesn’t nec­es­sar­ily lead to break­ing the monopoly? Maybe upon fur­ther in­ves­ti­ga­tion, Mon­santo doesn’t have a monopoly? Or there are mul­ti­ple or­ga­ni­za­tions who have copy­righted differ­ent gene ed­its so there’s no true monopoly?

The idea be­hind steel­man­ning soli­taire is to not stop at steel­man­ning the op­pos­ing view. It’s to steel­man the counter-counter-ar­gu­ment as well. As has been said by more elo­quent peo­ple than my­self, you can’t con­sider an ar­gu­ment and counter-ar­gu­ment and con­sider your­self a vir­tu­ous ra­tio­nal­ist. There are very long chains of counter^x ar­gu­ments, and you want to con­sider the steel­man of each of them. Don’t pick any side in ad­vance. Just com­mit to try­ing to find the true an­swer.

This is all well and good in prin­ci­ple but can be challeng­ing to keep it or­ga­nized. This is where Work­flowy comes in. Work­flowy al­lows you to have counter-ar­gu­ments nested un­der ar­gu­ments, counter-counter-ar­gu­ments nested un­der counter-ar­gu­ments, and so forth. That way you can zoom in and out and fo­cus on one par­tic­u­lar line of rea­son­ing, re­al­ize you’ve gone so deep you’ve lost the for­est for the trees, zoom out, and re­al­ize what trig­gered the con­sid­er­a­tion in the first place. It also al­lows you to quickly look at the main ar­gu­ments for and against. Here’s a worked ex­am­ple for a ques­tion.


That’s the broad-strokes ex­pla­na­tion of the method. Below, I’ll list a few poin­t­ers that I fol­low, though please do ex­per­i­ment and tweak. This is by no means a fi­nal product.

  • Name your ar­gu­ments. In­stead of just say­ing “we should buy or­ganic be­cause Mon­santo is form­ing a monopoly and mo­nop­o­lies can lead to abuses of power”, call it “monopoly ar­gu­ment” in bold at the front of the bul­let point then write the full ar­gu­ment in nor­mal font. Nam­ing ar­gu­ments con­denses the ar­gu­ment and gives you more cog­ni­tive workspace to play around with and also al­lows you to see your ar­gu­ments from a bird’s eye view.

  • In­sult your­self some­times. I usu­ally (always) make fun of my­self or my ar­gu­ments while us­ing this tech­nique, just be­cause it’s funny. Mak­ing your deep think­ing more en­joy­able makes you more likely to do it in­stead of putting it off for­ever, much like in­clud­ing a jelly bean in your vi­tamin reg­i­men to in­cen­tivize you to take that gi­ant gross pill you know you should take.

  • Mark ar­gu­ments as re­solved as they be­come re­solved. If you dive deep into an ar­gu­ment and come to the con­clu­sion that it’s not com­pel­ling, then mark it clearly as done. I write “rsv” at the be­gin­ning of the en­try to re­mind me, but you can use any­thing that will re­mind you that you’re no longer con­cerned with that ar­gu­ment. Fol­low up with a lit­tle note at the be­gin­ning of the thread giv­ing ei­ther a short ex­pla­na­tion de­tailing why it’s ruled out, or, ideally, just the named ar­gu­ment that beat it.

  • Pri­ori­tize rul­ing out ar­gu­ments. This is a good gen­eral ap­proach to life and one we use in our re­search at Char­ity En­trepreneur­ship. Try to find out as soon as pos­si­ble whether some­thing isn’t go­ing to work. Take a mo­ment when you’re think­ing of ar­gu­ments to think of the an­gles that are most likely to de­stroy some­thing quickly, then pri­ori­tize in­ves­ti­gat­ing those. That will al­low you to get through more ar­gu­ments faster, and thus, come to more cor­rect con­clu­sions over your life­time.

  • Start with the trig­ger. Start with a sec­tion where you de­scribe what trig­gered the thought. This can of­ten help you get to the true ques­tion you’re try­ing to an­swer. A huge trick to com­ing to cor­rect con­clu­sions is ask­ing the right ques­tions in the first place.

  • Use in spread­sheet de­ci­sion-mak­ing. If you’re us­ing the spread­sheet de­ci­sion-mak­ing sys­tem, then you can play steel­man soli­taire to help you fill in the cells com­par­ing differ­ent op­tions.

  • Use for de­ci­sions and prob­lem-solv­ing gen­er­ally. This method can be used for claims about how the uni­verse is, but it can also be ap­plied to de­ci­sion-mak­ing and prob­lem-solv­ing gen­er­ally. Just start with a prob­lem state­ment or de­ci­sion you’re con­tem­plat­ing, make a list of pos­si­ble solu­tions, then play steel­man soli­taire on those op­tions.


In sum­mary:

  • Steel­man soli­taire means steel­man­ning ar­gu­ments back and forth repeatedly

  • It helps with:

    • Com­ing to more cor­rect beliefs

    • Get­ting out of un­pro­duc­tive conversations

    • Mak­ing sure you do epistem­i­cally vir­tu­ous things that you already know you should do

  • The method to fol­low is to make a claim, make a steel­man against that claim, then a steel­man against that claim, and on and on un­til you can’t any­more or are con­vinced one way or the other