Encouragement to Instill Confidence?

I found this interesting and decided to overthink it.

Before proceeding, pause to think about whether this seems right to you and why.

The thesis here is something like:

  1. There is a right amount of confident to be.
  2. Encouragement proportionally raises confidence.
  3. (Implicit) Encouragement does not do other important things.
  4. →People often need encouragement.
  5. People tend to be less confident than they seem.
  6. Your estimate of their confidence will be based on how confident they seem.
  7. →You will tend to overestimate their confidence level.
  8. →You will tend to underestimate how much they need their confidence raised.
  9. Your estimate of how much encouragement raises confidence is unbiased.
  10. →You should give people more confidence than you expect them to need.

Also implicit is that ‘more’ is about the right level of adjustment here.

Once again: Does this seem right? Which steps seem right to you?

  1. There is a right amount of confident to be → Mostly True.

Confidence is strange. It is often a partially-self-fulfilling prophecy. It has various different effects. When founding a startup it can only be accurate at 0% but you will be better off with it higher, when looking to date similar dynamics often apply. Still, we can have one-dimensional ethics, and declare that of all the options there is almost always a best one. And under uncertainty, there is almost always a best one to push someone else towards.

  1. Encouragement proportionally raises confidence → True in expectation.

In general, encouraging someone will tend to raise their confidence. Some methods of encouragement do this more than others. Saying ‘that will work for sure’ will tend to focus its effects more on increasing confidence and in turn making action and hopefully success more likely. Saying ‘it’s good to attempt hard things, I’ll be proud of you for trying, good luck’ or ‘you will almost certainly fail or get turned down but the odds are so good you need to do it anyway’ might have different effects.

  1. (Implicit) Encouragement does not do other important things → False

Encouragement provides information, changes probability of various actions, changes social context and who can get or expect credit and blame from who, will effect credibility and so on. Encouragement is in part a claim that has a truth value, sharing knowledge and your guess on their chances of success and the wisdom and expected value of their actions. Often encouragement is not about confidence.

  1. →People often need encouragement → Mostly True.

This does seem true, but I often note that ‘need is a strong word.’ People often would benefit from encouragement. Often this is because they would benefit from more confidence, sometimes it is for other reasons. To say they ‘need’ encouragement is language I prefer not to use, but I mostly agree with the jist.

  1. People tend to be less confident than they seem → True by default.

A lot of naïve estimators are biased. This is one of those examples, mostly because a lot of people are continuously and intentionally faking confidence. There are exceptions and I know many of them, but mostly yes if you take someone’s confidence level ‘at face value’ you will guess too high. But guess what? If your estimates are known to be biased you should be adjusting them. It would shock me if Paul Graham hasn’t been making this adjustment automatically for a long time. He outright tells people, as advice, to present to people like him as having lots of confidence. I don’t make as large an adjustment because I interact with different people in different ways, but in many contexts I of course believe that how confident people present themselves and say they are are blatant lies. They, in turn, know everybody knows this, and this only increases their need to bluster to cancel out my downward adjustments.

  1. Your estimate of their confidence will be based on how confident they seem → True but incomplete.

I mean, yes, obviously, but it will also be based on how confident they could reasonably be in the situation, and how much incentive they have to present as overconfident or how much their circumstances would have trained them to be overconfident in this kind of situation and your guess on their view on distorting your perceptions of their confidence levels and so on. It all counts.

  1. →You will tend to overestimate their confidence level → Mu. That depends.

Again the obvious thing to do if this is true is adjust your estimate, rather than adjust what you do elsewhere to compensate. Doesn’t seem that hard. It’s not an especially difficult adjustment.

  1. →You will tend to underestimate how much they need their confidence raised → Mu because it depends on #7, also other factors.

If your estimates of confidence levels become unbiased the problem would go away, so this centrally depends on whether you have fixed that. If you haven’t, there is still the question of whether issues of confidence levels dominate. If your estimate of other impacts is unbiased and mostly continuous, it should be fine.

  1. Your estimate of how much encouragement raises confidence is unbiased. → Could go either way but my guess is more often false

My hunch is that most people underestimate the impact of their words of encouragement on confidence in many situations. Hearing such things often has a large impact, especially from someone you respect in the contextually important ways. It’s highly plausible this cancels out the original effect. In other situations, the lack of confidence runs deep, and the opposite effect can easily occur – ‘why aren’t you suddenly confident asking people out after I encouraged you?’ – but it does seem like we have a good sense that moving those dials is difficult.

  1. →You should give people more confidence than you expect them to need. → More often true and good advice than false and bad advice, but varies.

If nothing else this needs an ‘otherwise’ since it is an adjustment of a potentially biased estimate to remove the bias. Properly implementing it should also change your expectations and all that. In practice, doing more encouragement on the margin seems to be good on most margins where you would indeed prefer the thing you are encouraging. My guess to the primary mechanism there is more like ‘it would be mildly socially awkward and require doing/saying something/more rather than nothing/less which no one likes to do, so people do it less than they should.’ The ‘people are not as confident as they seem’ thing seems secondary to me, and if that was the only consideration I’d consider this mostly false. I also aim to focus more on the truth value of encouragement than the confidence effect, for both some good reasons and some personal preference reasons.

Recently I did offer people encouragement, in particular to apply for SFF funding. How much of this was because I was estimating that (1) charities lacked confidence in their chances of success and (2) they lacked confidence in their ability to ‘do it right’ as opposed to having a lack of the required knowledge? Somewhat, but also because my gut told me they were also (3) unaware of the opportunity, (4) lacked the required knowledge which I attempted to provide, and (5) treated the cost of applying as higher than it is and treated failure as far more painful than it would be. When such folks ask me more questions directly, as several have, I am careful only to offer careful and directed encouragement where it makes sense.

My overall guess is that there is too much and overly intense encouragement that is not genuine, as in not done because one feels the actions in question would be worthwhile. There is too little and insufficiently intense encouragement that is genuine, with the need to further raise confidence levels a substantial portion of that but not dominating other reasons especially trivial inconvenience, lack of motivation and mild social awkwardness.

This entry was posted in Uncategorized. Bookmark the permalink.

23 Responses to Encouragement to Instill Confidence?

  1. Sam says:

    I’m not sure if this is directly relevant to the context of Paul Graham’s tweet, but I feel that teachers and mentors are often stingy with praise & encouragement (or they overcompensate, and give too much worthless praise/encouragement). When I make this mistake myself, I often assume that the person at the receiving end knows they did a good job and thus doesn’t need me to tell them. After one of my graduate students gives a talk, for instance, my natural inclination is to tell them everything they did badly and let them figure out that the rest was good.
    One of the other reasons is that it can feel presumptuous to praise someone (or encourage them) — it asserts you are their peer (or outrank them) and that you are qualified to give an opinion.

    • TheZvi says:

      I think this is basically right – such folks often don’t consider this part of their jobs at all and people get the idea that they suck when they very much don’t suck, or have no idea where they are at. But I don’t think this is a need-more-confidence problem so much as I-have-no-idea-what-message-this-is-sending-or-that-I-ever-need-to-do-this thing. Still, important thing to remember.

      The ‘not a peer if you praise them’ thing is… weird. I praise peers!

      • Sam says:

        Sorry, I was unclear. I find it slightly inappropriate to praise someone who outranks me. If I praise someone, I am asserting that I am at least at their level. I would find it slightly weird to praise my department chair about something he did, and incredibly difficult to praise my former PhD supervisor for anything mathematical (I’d praise his bbq ribs no problem). [For context, I am a mid-career math professor.]

        • Jon Task says:

          A related thing is a point in Japanese (and more traditional circles of Chinese) culture, praising somebody at / above your level can be seen as a dreadful insult. We have norms where only your superior is allowed to evaluate your performance, and praising somebody indicates that you see yourself as their superior.

    • waltonmath says:

      Glad to see someone reporting this phenomenon! This resonated with me, and still puzzles me. I’m not even sure if just giving criticism is a mistake — good criticism will help someone make good changes, and for some people in some situations, pure criticism might be the most efficient feedback you can give. But it’s possible someone will engage with criticism as “that person thinks I’m doomed to suck” or some such, in which case it could be counterproductive.

      I feel like I’m more likely to praise something if it makes me enthusiastic. I guess I’ve sometimes gotten confused thinking that I’m supposed to like everything that students, or other types of people, say, but not all of me actually agrees with this perspective, which gives weird results.

  2. Greg kai says:

    Hum. Many people are not confident enough (usually pessimistic and too risk-averse), meaning they will fail less than they think and encouraging them is a net gain even if you don’t have the faintest idea about their chance of success yourself…
    But some are over confident (optimistic, failure is always a surprise and explained by hostile/unfair behavior of competitors or judges, not lack capacities or even bad luck). I know some of those people, they are energetic and entertaining but also tiring and dangerous associates for stuff you can not afford to fail at.
    I suspect they are as numerous as the first type on the whole, but probably not in the same circles : in mine they certainly are a minority.
    So it really depend on who you talk, on particular their ability to self evaluate and their level of optimism. I would encourage only pessimists with not so great self evaluation (where you can both evaluate better than they do their chances of success and make them less afraid to try what it’s worth trying.), but this particular combination is not so frequent, at least in adulthood… In particular, you better be better at evaluating success chances and failure prices than they are, because else it’s typically “just try it, you have everything to win and (i have) nothing to lose”

    • Greg kai says:

      But that ‘s before trying. After the fact, it’s always nice to give an honest but gentle appreciation. As long as you don’t nag and honestly say that’s your opinion only, it help self calibrate. But that’s appreciation, not encouragement, at least that’s how u understand it… And being too nice at the price of honesty decalibrate, so imho it’s not really a good thing to do…

    • TheZvi says:

      I’d be very surprised if the overconfident were not very outnumbered. They do exist but in my experience they are pretty rare.

      • Tobias Pace says:

        Depends on the field. Most high school basketball players are massively overconfident about their chances of making the (W)NBA. Same goes for many musicians, actors, and fiction writers. In these fields and many others, the rock saying “You won’t get rich doing this” will be an extremely accurate predictor.

        • Greg kai says:

          Exactly my point. I am in engineering (the more academic side of it), and there indeed overconfidence is not the norm (although it’s sometimes hard to tell, because it is often faked especially at the managerial/commercial levels).
          But in other fields, it’s not the case. In fact, all high risk high rewards circles should be filled by overconfident types (a few successful, most not and getting bitter as they age), by definition.
          People with very efficient safety net also seems overconfident, because they do not support the full cost of their failures. Are they really? Not sure, at least not until their safety net fails…

        • TheZvi says:

          Yeah, there are what we might call “rock star” systems where the odds are HORRIBLE, so almost everyone who actually tries is overconfident (startups, writers, competitors of most sorts, etc) due to selection effects.

      • My gut is that people are probably socially underconfident. But I’d guess they’re epistemically overconfident.

  3. cherisium says:

    I find your discussion of point 4 strange. If I’m understanding this correctly, you came up with the chain of reasoning and its wording. So why then spend point 4 on the distinction between “need” and “benefit from”? If it’s language you prefer not to use, why did you use it in the first place? I found this point sufficiently confusing that it distracted me a bit from the rest of the piece.

    • TheZvi says:

      Because Graham uses the word ‘need’ in exactly that way. Interesting that it was still off-putting. Should I have noted this explicitly, or something else? Curious to hear your thoughts.

      • Sawyer says:

        It’s funny, I had the same distraction, scrolled up, and then realized it must’ve been in the thread (but didn’t bother to actually click through). But small data point that it might be worth noting explicitly.

  4. 5hout says:

    In a work context I am required to frequently give feedback to teams I am placed over. The teams are assembled ad hoc for projects and change often. Average project duration is probably 5 weeks, but with a high variance (i.e. many projects are 10 days and a handful are 8-12 months). I have been fully remote for ~5 years, but previously did this work in person.

    In person average people got very little genuine positive explicit feedback, but had positive normal social interactions. One thing I have noticed since going remote was that average people were still getting little genuine positive explicit feedback (maybe, 1/10 projects you’d get a little personalized note or something), but had lost out on all the positive normal social interactions. This left only genuine negative explicit feedback (given on a daily/weekly basis as needed), and the implicit genuine positive feedback of “Large pool of applicants for each project, so continued employment (or movement to quality control team) means you are doing a good job”.

    With the increase of people going remote these last 2 years I’ve seen this lack of genuine positive explicit feedback start to take a toll (hypothesis: strong self selection/filter effect before the pandemic meaning the fully remote people of 3+ years are mostly odd people with different motivations) and people have gone off the rails a bit. I have tried to give such feedback, but only as appropriate (i.e. not adjusting for an estimate of their “needs” or other biases, simply “are they doing a good job”).

    The response has been… weird? Very cringey to be honest. Sufficiently annoying that I have sort of phased it out. Maybe the over-tuned response is a sign people need more such feedback, but it was problematically weird. Notes from people under me to my boss/bosses boss about how great it was to get positive feedback. People DM’ing/emailing me personally for assignments (I have very little control of this and it’s a massive nono). People saying awkward/cringe stuff in chat about getting feedback.

    Probably does mean they desire this feedback more, but it was enough of a problem for me that I only do it on very rare occasions now, and have returned to primary genuine positive feedback mechanism being related to rating people higher at end of projects so they get hired again.

  5. waltonmath says:

    Wrote up some reflections in response to your “think about it first”s, and then in response to some stuff in the rest of the post: https://docs.google.com/document/d/1gMgwx6q6tTGPNz_jI3aQFoj0hxCDZ9xQz70Uc0PXLGU/

  6. Ninety-Three says:

    “When founding a startup it can only be accurate at 0%”
    Wait what? If I understand you correctly, you’re claiming that not only has no one ever accurately assessed their startup’s chances, but for some reason this isn’t even possible? Even supposing that there exists no human on the planet who can conclude “we’re only 20% to succeed” without that discouraging them so much as to lower their chances below 20%, there has to be some number high enough that it’s no longer discouraging to find, and some scenario where someone accurately predicts that number.

    • Ninety-Three says:

      Imagine that I edited post my to read “no ever accurately assessed their startup’s *nonzero* chances”.

      • TheZvi says:

        The model is that fully thinking one’s chances are low hurts one’s chances sufficiently that someone who thinks they are unlikely to succeed never will – the odds are always low, so if you say 20% then you’re closer but it’s not going to be 20%. I don’t know if this is fully true but it is effectively conventional wisdom among VCs and founders, and if you are this pessimistic and also unwilling to lie about your view of your chances you won’t get funded.

  7. bugsbycarlin says:

    More dakka.

  8. cakridge2 says:

    > Saying ‘that will work for sure’ will tend to focus its effects more on increasing confidence and in turn making action and hopefully success more likely. Saying ‘it’s good to attempt hard things, I’ll be proud of you for trying, good luck’ or ‘you will almost certainly fail or get turned down but the odds are so good you need to do it anyway’ might have different effects.

    Maybe I’m kinda strange, but I’d rather hear the latter over the former – saying “that will work for sure” is a nice thought, but you don’t know that for sure. The second/third statements are much closer to the truth.

    • TheZvi says:

      There is a group of people, who I like to be around as often as possible, who think this way or at least think they think this way. They are in the minority for sure.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s