Why do we need a separate site for Vi/Vim?

Question

Proposal: Vi and vim

Vim already falls in the scope of at leat SuperUser & StackOverflow; who do we need a separate StackExcange site?

Answer 1

Number of questions with the tag Vim & Vi:

StackOverflow: 15260, 1126
SuperUser: 2624, 181
UNIX: 742, 164
ServerFault: 31, 79
Programmers: 23, 5
TeX: 164, 0

There are at least 3 different places with a significant number of Vim/Vi related questions, and 3 sites with some questions. There are probably some other sites with Vim/Vi related questions as well.

This the same reason why an Emacs, TeX, or Ubuntu SE exists, all 3 SE sites fall within on the border between StackOverflow & SuperUser, and questions are scattered over those 2 (and some more) sites. Having a single resource would be an added advantage.

Answer 2

Just a copy/paste of my answer from the previous discussion:

I'd say that the strongest argument we have in favor of a separate Vim site is that Vim Q&A is already quite fragmented across the entire network.

• Stack Overflow: 15221
• Super User: 2671
• Unix & Linux: 790
• Ask Ubuntu: 373
• And even TeX - LaTeX with 156

First of all, almost any question at all about Vim can fit on either Stack Overflow or Super User. Indeed, I've searched for Vim questions on SO before only to finally find an answer on SU. This is a Bad Thing™.

Furthermore, fragmentation means you're not reaching your full potential audience. What if I have a question about Vim-LaTeX on Ubuntu? I only eventually decided to post on Ask Ubuntu because it got more traffic. This is also a very Bad Thing™. The people over on Tex - LaTeX don't get to see my question, so it gets less of a (specialized) audience. People from TeX - LaTeX might have the same problem, but give up because they can't find it on that site. Less questions + less answers = less happy unicorns.

Also, it's not as if I'm the first to bring this up. But yeah, any argument you can make against Vim hurts your Emacs site, and anything good you say about those Emacs people over there applies equally to us too.

Answer 3

There was a large (now deleted) discussion of this on the old Vi(m) proposal. The closing of the proposal resulted in the most downvoted announcement on Area 51 (this announcement has also been deleted), with many answers arguing against the closing.

Still, the motivation for closing a fully committed proposal remains a mystery for me, but I would like to quote the reasoning from this answer, which discusses why the Coffee proposal might be granted a site of its own, although there is already "Seasoned Advice" for "Food & Cooking experts"

A network (or group of sites) isn't always worth exactly the sum of its parts. If there is an audience out there we are failing to attract by lumping all these subjects under one umbrella (a culinary site), then splitting that subject off into a site of its own may actually result in a net gain overall. But if we're simply splitting apart one site to create two — for, essentially, the same audience — then that would be a poor allocation of subject space and completely undesirable… and we would take appropriate action.

But the only way to know for sure is to give the site a try. And so far the subject of "Coffee" has passed the first trials of Area 51: gauging interest.

It seems to me that the same reasoning may apply to granting Vi(m) a (try at a) site of its own, since I think there exists a different audience for Vi(m), which doesn't fit under the umbrella of StackOverflow users. (I assume that this reasoning also supported the creation of the Emacs proposal.)

Answer 4

Obligatory caveat: I haven't discussed this with the rest of the Community team. The opinions below are my own and I reserve the right to change them without notice. _{Satisfaction guaranteed or double your money back. (Offer void for internet transactions.)}

I think the best site for Vi/Vim questions is Stack Overflow. If you are already interested in asking and answering questions about this family of editor, you really owe it to yourself to follow that filter for a few days. I just can't help comparing these actual questions to the example questions on this proposal. Area 51 example questions rarely make good questions on beta sites, but many of the examples here seem perfectly topical on Stack Overflow.

To tackle the elephant in the room: Emacs set the precedent for splitting off programers' editors from Stack Overflow. Before the private beta, I proposed some criteria for spinning off topics. A topic split can only be successful if the new site reaches a new audience; cannibalizing an existing site is both inefficient and detrimental to people seeking answers. TeX/LaTeX was a good candidate because so many TeX users (mathematicians) do not consider themselves programmers. The same can be said for the various CMS sites.

After 122 days in beta, we have enough data to see if the Emacs spinoff was successful. Before the private beta, there were ~7 Emacs questions per day across the network. As of today, the rate is ~5 q/day with the emacs tag. However, Emacs.SE is getting nearly 12 per day. Something like 40% of those questions (by my estimation) would be off-topic on any other site on the network. A reasonable guess is that splitting off the topic added somewhere between 3 to 10 Emacs questions a day to the network.

Do those questions get answered as quickly and as well as they would on Stack Overflow? The data suggests they are:

Tag/Site         questions views score closed answers accepted answer TTA
                          median   avg      %     avg      avg      %   *
-------------------  ----- ----- ----- ------ ------- -------- ------ --- 
[elisp]               3088   165   3.2    1.8     1.9     77.3   96.8  56         
[emacs]              12161   234   3.7    3.3     1.9     68.2   93    70
Emacs                 1564    77   4.3    1.6     1.6     65.3   91.7  69 

It turns out that the [emacs] tag is not particularly well-served compared to other tags in terms of time-to-answer (TTA):

Stack Overflow     8684846   243   1.6    4.4     1.7     57.1   88.3  24

Which brings us to the topic at hand:

[vi]                  1157   275  11.3    6.7     2.8     71.5   95.9  10         
[vim]                15642   254   5.2    4.4     2.1     73.1   95.8  23          

Questions in the [vim] tag are answered as quickly, more often, and better than the typical tag. Plus those answers are viewed by more people. The data strongly suggests that Vi/Vim askers get the answers they need on SO. I'd be shocked if Vi.SE could improve on that.

However, I'm willing to give this proposal a shot at private beta. Partially, I'm encouraged by the momentum of commitment:

As I mentioned on the Emacs proposal:

The first couple of pages of committers by reputation shows these are folks who are very comfortable with our network, but still want a [dedicated site]. These are people who are voting with their feet. It seems very likely that our existing sites are not meeting their particular needs.

Like the Emacs committers, a good percentage are active on other sites:

committers active in
65.4% Stack Overflow
24.1% Super User
20.3% Unix & Linux
18.0% Meta Stack Overflow
17.3% Meta Stack Exchange
15.8% Ask Ubuntu

I still think this proposal is a waste of time and effort. Maybe I'm close-minded and biased, but I don't suspect that a site for Vi will draw a different audience than the one we already have on Stack Overflow. There will no doubt be plenty of questions in the private beta, but I'm going to be looking for questions that can't be asked on other sites. If Vi.SE succeeds, it will be by addressing questions that have nothing to do with programming.