[Neuroimaging] When to use neurostars, and when not to

Matthew Brett matthew.brett at gmail.com
Thu Aug 27 10:16:33 CEST 2015


On Wed, Aug 26, 2015 at 11:17 PM, Eleftherios Garyfallidis
<garyfallidis at gmail.com> wrote:
> Hi all,
> My initial personal point of view was to use one mailing list both for users
> and developers. That is neuroimaging at python.org (previously nipy-devel...)
> Then I saw that this list was not attracting enough questions. This is
> probably because we didn't promote this list enough. And maybe that
> should have been the first thing to do. Which we clearly didn't do.

Of course, there will always be a spectrum of users, from "just tell
me how to use it without it giving me an error message" to "I want to
write new stuff for the package".

We were discussing this on the phone yesterday, but for me the
question is - what do we hold as the model for users in general?  Do
we hold that all users should expect to contribute to the discussion
of how to build and maintain the package, or do we think that some
users can be safely left to - just use the package.

As you know - I think we're running into severe trouble in
neuroimaging, as the software becomes easier to use, and there are
more methods available, without any matching increase in training.
So we have more and more people using software that they do not
understand.  Inevitably, this means more bad science.

The only way to avoid that, it to change the culture.  We want to
switch from the classic model of "we are experts, trust us, use the
software" to "we are all learning how to do this.  We are learning by
building. Come join in this difficult task".

This is why I feel so strongly that we should not separate user and
developer questions.

> Then I thought okay the neuroimaging at python.org list is for developers and
> expert users and neurostars.org
> is for users who don't want to touch development and need feedback.

Hence - I want to say to these type of users - "this is not a model we
support.  Science is the disbelief in the opinion of experts.  We
won't stop you doing this, but neither will we imply this is a way of
working that we encourage or expect."

> But I have to say that there are some features in neurostars which can be
> dangerous. One of them is that people who answer questions get
> different status and reputation points. For example, the status of the
> person answering often says "trusted". This is misleading information.
> I think we need to be smarter here. There are no experts. We are research
> projects. We want a community which does not promote
> expertise but promotes understanding and good scientific practice. The focus
> of neurostars should be exactly that.
> I see that neurostars is still in a beta and I am wondering if we can change
> some of its settings.
> It is true that neurostars has attracted very quickly a relatively large
> audience. So, I am not sure if we should stop using it or not. But I will
> suggest something different.
> Can we continue using Neurostars with some different setup so that it
> promotes more understanding/discussion rather than expertise?
> To be more clear. Here is a summary of the important problems that
> StackOverflow has which also Neurostars can potentially have:
> This is from
> http://michael.richter.name/blogs/awhy-i-no-longer-contribute-to-stackoverflow
> which was posted previously in this thread. Thanx for
> posting this.
> The problems with StackOverflow are summarized in this list:
> Poor pedagogy
> Poor reward system
> Poor community

I'm afraid I believe these features are structural to the
stackoverflow "approved answer" model.   What we need is a model that
encourages discussion, and the expectation that the beginner should
expect to become a peer.  This is what mailing lists are very good at,
and approved answer models are very bad at.

See you,


More information about the Neuroimaging mailing list