Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Don't use the mailing lists

blog.ironholds.org

Post reply on HN
Print view View original

1–10 of 34 posts

Re: Don't use the mailing lists

#3

Post by keithpeter »

CRAN is core and has to be trusted, the alternative community pointed to by OA provides a way of publishing newer libraries and applications built on core.

Seems logical to me that the barrier to the former is higher than the barrier to the second (essentially no barrier, just put it out there and see if people like it and be prepared to respond to feedback).

Perhaps the 'tone' of that higher barrier to the core could be adjusted and mechanisms for submission made clearer, but I think that there needs to still be a standard.

Am I wrong?

Re: Don't use the mailing lists

#4

Post by edtechdev »

The same thing seems to happen in many mailing lists or newsgroups once they reach a certain size, as well as on discussion boards, stackoverflow, and other sites like wikipedia. One person called them 'RTFM jerks': http://www.linuxtoday.com/infrastructure/2006033100126OPCY On the python lists, they were called NIMPY ("Not in my Python!"): http://www.artima.com/weblogs/viewpost.jsp?thread=87182

You combine a power structure (control over others in some respect) with pseudonymity and you'll accelerate this problem. Even without a power structure, people will try to dominate, bully others, by posting more frequently than anyone else in an effort to drown out other opinions or contributions. The anonymity disinhibits this behavior further (called the "greater internet fuckwad theory"): https://en.wikipedia.org/wiki/Online_disinhibition_effect

Re: Don't use the mailing lists

#5

Post by mziel »

Ultimately open source software is approved by many eyes, for which Github seems like a natural place. Many excellent packages are already only available via devtools::install_github(). CRAN is nice and easy (especially for people starting with R or with programming) but is not a necessity.

Re: Don't use the mailing lists

#6

Post by avivo »

Would one potential solution be to create a secondary optional gatekeeper?

The basic idea would be - don't submit directly. First submit to the community gatekeepers, they will look over your work, and if they think it will meet approval, they'll pass it on. Otherwise they can give helpful feedback. They can have strong standards about how contributors are treated.

This extra step can be optional - if you don't mind rough treatment or are confident of approval you can skip it...

The result: less work for existing gatekeeper(s) without changing their process, more friendly and helpful interactions for contributors, and no slowdown if you really know what you are doing.

Of course the cost is that people need to volunteer to be that secondary gatekeeper! (and have some tough skin when interfacing with the primary gatekeeper)

==

In an ideal world everyone would have the skills and interest to efficiently be pleasant to everyone else while vetting infinite code. But in this world of finite time and not everyone having every skill it can make sense to have some more "division of labor".

That said - I'm very curious if there are aspects of this that I'm missing that make this perspective insufficient.

Re: Don't use the mailing lists

#7

Post by _delirium »

mziel wrote: post #5

Ultimately open source software is approved by many eyes, for which Github seems like a natural place. Many excellent packages are already only available via devtools::install_github(). CRAN is nice and easy (especially for people starting with R or with programming) but is not a necessity.

As a user I'm a bit wary of installing packages directly from GitHub. The CRAN review process isn't perfect, but it's much less likely that some malicious code accidentally slipped in, or that someone is namesquatting on a popular package name but is not the genuine package. With a random GitHub link I'd feel like I need to do this reviewing work every time myself: investigate who owns the GitHub repository, what their track record is, what recent commits there are and by whom and how reviewed, etc. That's a lot more work than using the heuristic that "CRAN code is probably safe". More mundanely, CRAN typically does a good job with dependencies, and is usually in a working state.

I could see alternate curated repositories from CRAN, but to switch from CRAN I'd still want some kind of curation.

Re: Don't use the mailing lists

#8

Post by zby »

edtechdev wrote: post #4

The same thing seems to happen in many mailing lists or newsgroups once they reach a certain size, as well as on discussion boards, stackoverflow, and other sites like wikipedia. One person called them 'RTFM jerks': http://www.linuxtoday.com/infrastructure/2006033100126OPCY On the python lists, they were called NIMPY ("Not in my Python!"): http://www.artima.com/weblogs/viewpost.jsp?thread=87182 You combine a power st…

"Even without a power structure, people will try to dominate, bully others, by posting more frequently than anyone else in an effort to drown out other opinions or contributions."

This can be easily countered with setting user quota (per day of week or whatever) on emails to the list. Not that this is a cure for all - but I believe this can improve the democracy of mailing lists. I hope some mailing list management software will add this as an option some day.

Post reply on HN