Hacker News Reader

Terse feedback can be fine, once you know that, for a specific person, terse feedback is effective.

I have met more than a few people that said they preferred "brutal honesty", but would still react very defensively to straightforward, terse feedback.

Being terse without seeming to be on the attack is very difficult, and it is human nature to defend ourselves -- and stop listening! -- if we perceive that we are being attacked.

> I don't need to hear compliments on what's right... just what can be improved.

I used to say the same thing, but now I think I was wrong. I didn't want to hear compliments on what's right. But I did need it. Focusing on negatives helped me reduce error rates. But ignoring the positives limited my ability to grow.

What got me to slowly switch was learning about appreciative inquiry. It's a practice of focusing strongly on the good aspects of something and seeking to amplify them. I looked into it because a team I was working with had these brutally negative retrospectives. They were in a rut of flagellating themselves every week about all their failures. They were a great team, but weren't performing well because they became gradually more stressed and discouraged. And after I saw that in them, I eventually saw it in myself, too.

I'm now 6 or 7 years into consciously switching approaches and I'd never go back. Just today I was pairing with a younger programmer, and we even took some time to appreciate their anxiety. That fear of failing was causing them a lot of problems, but instead of asking them not to worry, I asked them to channel their anxiety into areas where it was useful, like writing good tests.

That said, I'm with you on not wanting "nice" versions of things, which I see as a social performance, and not actual kindness.

> for type of people who appreciate receiving it

This is key. I wonder though, what's a good way to signal that to friends and coworkers? I personally declare Crocker's rules[0], but I rarely have the opportunity to communicate that to people I'm talking to.

[0] - https://wiki.lesswrong.com/wiki/Crocker's_rules

"By declaring commitment to Crocker's rules, one authorizes other debaters to optimize their messages for information, even when this entails that emotional feelings will be disregarded. This means that you have accepted full responsibility for the operation of your own mind, so that if you're offended, it's your own fault. The underlying assumption is that rudeness is sometimes necessary for effective conveyance of information, if only to signal a lack of patience or tolerance: after all, knowing whether the speaker is becoming angry or despondent is useful rational evidence. Two people using Crocker's Rules should be able to communicate all relevant information in the minimum amount of time, without paraphrasing or social formatting.

Thus, one who has committed to these rules largely gives up the right to complain about emotional provocation, "flaming", "trolling", "abuse" (hopelessly subjective terms) and other alleged violations of etiquette. They give these rights up in the interest of effective debate."

I don't think being blunt is more efficient. First, adding a few "kind" softener words do not really take so much time, don't you think so? Second, efficiency in human to human communication is not only a function of raw information density. Adding well chosen "kind words" may allow the reader to get some context and a better understanding of the message.

A simple example: say you already communicated with the guy in the past. If you add a reference to this previous communication, the reader will remember you better and adjust the frame.

There's so many things passing through language, many are very volatile but still important. We're not computers. Human language have very little in common with computer languages.

I think it's already captured in Crocker's rules definition - "one authorizes other debaters to optimize their messages for information, even when this entails that emotional feelings will be disregarded". It's not about being rude for the sake of being rude - if a kind word provides some context best, a kind word should be used. The idea here is to free your interlocutor from having to constantly censor their communication in order not to accidentally offend you.

Interesting, never heard of Crocker's rules before. Fascinating, and there are parts of it I definitely like, especially the bits around "accepting full responsibility for the operation of your own mind."

That said, this assumes that all communication is between two individuals, and doesn't provide signaling to other members of the group. For instance, if both you and I adopt crocker's rules, but Bob and Susan do not, a new member of our group may assume incorrectly that this type of communication is appropriate for Bob and Susan, or create an environment where the new/less powerful member may be less willing to speak up because of the perceived tone of communication.

It also seems to be position of relative power: one adopting Crocker's rules assumes to be in a position where certain types of speech do no harm to them.

I think one of the important takeaways from the talk is that even if people say they want brutal feedback they probably don't. More importantly, feedback can be totally honest without compromising it's optimism.

> I would much rather get a code review of "Missing tests for negative numbers, empty string, null" than a longwinded and "nice" version

I agree that niceness isn't always necessary. Fortunately, that's not what the talk is about. For me, the examples you gave actually fulfills all the criteria for optimistic criticism:

- It's impersonal: the focus is on what the code is missing, not your ability as a coder.

- It's specific: there are hyper-specific criticism.

- It's temporary: the tests are merely "missing" and that's a condition that can be quickly changed.

This is a totally valid point about my example. Unfortunately for the sake of my argument, it was done by accident (or subconscious).

I have been practicing detaching critical feedback from the person and instead attaching it onto the situation or context for quite a while now. In any form, I think that point is crucial.

>I would much rather get a code review of "Missing tests for negative numbers, empty string, null" than a longwinded and "nice" version.

Could you give an example of what you're imagining a longwinded nice version would be like? Because this seems mostly fine, just a bit clipped.

All I'd really expect to improve it is something like: "This is mostly fine. It just needs tests for negative numbers, empty strings and null." Adding just a few words validates their work, and more practically, it also clarifies that the criticisms are exhaustive.

The structure of what I had in mind was going through a laundry list of aspects of code (functionality, style, structure, functional decomposition, algorithmic efficiency, test coverage, use of libraries, blah blah blah), and making specific positive comments about the things that are already good. Personally I don't glean much value from that i.e., the compliments aren't something I'm concerned with and the feedback isn't something I can act on anyway.

OTOH, by going through that laundry list, it shows that you actually performed a full review. You can optimize that (for example, by having a checklist where you mark off all areas with either constructive criticisms or a checkmark.

>> When giving feedback, some people say that they are brutally honest

> As someone who prefers giving and receiving terse feedback... I would suggest that there's nothing wrong with brutal honesty for type of people who appreciate receiving it.

It seems to me that you cited the fragment of donw's comment about "brutally honest", but ineffective feedback ("The module you wrote is crap because your code sucks and is hard to maintain.") and then started arguing that some people like receiving terse feedback.

But won't you agree that the example of "brutally honest" feedback wasn't in any way terse and that "brutally honest" doesn't necessarily mean terse?

I find the second part of the "kind" feedback much more terse: "Breaking those big methods and classes into smaller separate objects and functions would help a lot".