librelist archives

« back to archive

Meta-nitpicking

Meta-nitpicking

From:
Mark IJbema
Date:
2013-09-21 @ 16:58
One of the problems exercism currently have is that not all feedback is of
good enough quality. Can we think of a way to improve this? Some ideas:

Better instructions (I think the current how to nitpick is great, but I
think it would work better if it wasn't just a big piece of text)

Allow for meta-feedback which is less obtrusive (what if you could
meta-nitpick a nitpick, and it shows only to the nitpicker?)

Rate feedback (this was suggested elsewhere, not sure how it would tell
people how to improve though)

Any more thought?

Mark

Re: [exercism] Meta-nitpicking

From:
Katrina Owen
Date:
2013-09-21 @ 19:59
On 09/21, Mark IJbema wrote:
> One of the problems exercism currently have is that not all feedback is of
> good enough quality. Can we think of a way to improve this?

Thank you for kicking off this discussion!


> Some ideas:
> 
> Better instructions (I think the current how to nitpick is great, but I
> think it would work better if it wasn't just a big piece of text)
> 
> Allow for meta-feedback which is less obtrusive (what if you could
> meta-nitpick a nitpick, and it shows only to the nitpicker?)

In many ways I would prefer to focus on better instructions rather than
meta-feedback, because if it is done right, we won't need the
meta-feedback other than perhaps a discreet pointer towards the
instructions that are there.

The problem that I'm struggling with right now is both communication and
UX. I have no idea where to put things so that people see them at the
right time, or how to structure things so that they make sense.

> Rate feedback (this was suggested elsewhere, not sure how it would tell
> people how to improve though)

Eh. Yeah, I'm not convinced this is a way forward -- in particular
because it doesn't tell you what you did well/badly or what you should
be striving for.

In summary, I'd love to focus on the first idea: How to provide better
instructions, or how to provide the right instructions at the right
time.

I did a write-up for the "typical ruby bob":

http://exercism.io/help/nitpick/ruby/bob

I haven't linked to it from anywhere yet, because it's just a blob of
text, and I'm not quite sure whether it's going to be useful to anyone.

Katrina


Re: [exercism] Meta-nitpicking

From:
Barrett Clark
Date:
2013-09-21 @ 21:02
I like the idea if giving some more structured and detailed feedback. I 
think the Bob text is good, too. I've found that some people are really 
interested in the feedback loop and the process of discovering new ways 
and approaches. Some people are also just trying to learn a new language 
with easily digestible sample problems. The feedback loop is especially 
great for them.

And then there is another group of people who argue with or otherwise 
disregard feedback. Disagreeing with feedback is fine, but I've left 
comments once or twice to the effect of "we take time to give thoughtful 
feedback for a reason."

Upvoting and those sorts of things are necessary in an ecosystem like 
Stack Overflow. You get into ranks and crap like that too. I think there 
is definitely feedback that is less helpful, and also lots of room for 
discussion. That's one of the great things about exercism in my mind. 
Great exercises and feedback loop with peers/smart people.

~Barrett

> On Sep 21, 2013, at 2:59 PM, Katrina Owen <_@kytrinyx.com> wrote:
> 
> I did a write-up for the "typical ruby bob":
> 
> http://exercism.io/help/nitpick/ruby/bob
> 
> I haven't linked to it from anywhere yet, because it's just a blob of
> text, and I'm not quite sure whether it's going to be useful to anyone.

Re: [exercism] Meta-nitpicking

From:
Katrina Owen
Date:
2013-09-22 @ 01:51
On 09/21, Barrett Clark wrote:
> I've found that some people are really interested in the feedback loop 
and the process of discovering new ways and approaches. Some people are 
also just trying to learn a new language with easily digestible sample 
problems. The feedback loop is especially great for them.

I hadn't thought about it in terms of separate groups. While it might be
a bit of a false dichotomy, I think it is probably useful to think of
them as separate groups.

> And then there is another group of people who argue with or otherwise 
disregard feedback. Disagreeing with feedback is fine, but I've left 
comments once or twice to the effect of "we take time to give thoughtful 
feedback for a reason."

Sometimes people ignore everything I've suggested without any comment,
and resubmit things with exactly the same patterns and things that I've
commented on.

I find that pretty maddening. Not because I'm right, but because they
haven't even taken the time to tell me what they're thinking. Occasionally
I'll say something like "Clearly you disagree, would you mind helping me 
understand why?".

> I think there is definitely feedback that is less helpful, and also lots
of room for discussion. That's one of the great things about exercism in 
my mind. Great exercises and feedback loop with peers/smart people.

Yes, I can point to specific feedback that is less than helpful.

For example, feedback that is based only on speculation about the
future: "You should extract a class because it will make it easier for
you to add other types of objects later". I find this to be less than
helpful. I'd like the feedback to be about the code itself: You have
three private methods that all take the same parameter. Perhaps there's
an abstraction hiding there.

Or feedback that basically just says 'Nice work' or 'Good job'. What is
good about it? What is nice? Are there expressive variable names? Good
separation of concern? No duplication? An elegant turn of phrase? How am
I going to reproduce the _goodness_ factor in other code if we don't
have a conversation about what exactly makes it good?

Or feedback about itty-bitty details when there are glaring big-picture
concerns to talk about.

I feel like there should be some way of helping people understand how to
provide better constructive criticism and specific praise, but I can't
think of any reasonable approaches to this.

Katrina

Meta-nitpicking

From:
Jason W. May
Date:
2013-09-30 @ 20:30
(I just joined the list, and there's no way to Reply to an existing thread.)

I'm interested in ways to move interesting conversations out of individual
nitpick sessions into a more general forum. But it should still be somehow
connected to Exercism and probably to that particular exercise, to provide
context. I like the idea of restricting participation in some manner - 
perhaps to Exercism users who have completed that exercise - to keep the 
chatter on topic and limited to folks who have made at least a minimal 
investment in the process.

Some of the recently-added exercism features ("OK, I'm done" and 
encouraging multiple nitpickers) help to keep individual exercises from 
getting bogged down in unhelpful back-and-forth.  But there's often value 
in having those conversations, once they aren't blocking progress.

I don't have any firm ideas on how such a forum might be implemented. A 
Discourse instance parallel with exercism is easy to set up, but that 
doesn't do anything to help encourage conversation to happen in a manner 
that participants can find easily.

-Jason

Re: [exercism] Meta-nitpicking

From:
Katrina Owen
Date:
2013-10-01 @ 01:28
On 09/30, Jason W. May wrote:
> I'm interested in ways to move interesting conversations out of 
individual nitpick sessions into a more general forum. But it should still
be somehow connected to Exercism and probably to that particular exercise,
to provide context.
<snip>
> I don't have any firm ideas on how such a forum might be implemented. A 
Discourse instance parallel with exercism is easy to set up, but that 
doesn't do anything to help encourage conversation to happen in a manner 
that participants can find easily.

I would love to have these meta-discussions. I don't know if you saw
something in an earlier discussion here, but I felt the need to do a
meta-analysis of the typical "bob", and had nowhere good to put it, so I
ended up putting it on the website, but I haven't linked to it anywhere.

http://exercism.io/help/nitpick/ruby/bob

I think I need to pay a UX person to help me think through this muddle
of ideas and desires. I can't figure out where to start.

Katrina