Ask HN: What's the deal with mailing lists? | Hacker News Reader

Ask HN: What's the deal with mailing lists?

29 pointscool-RR34 comments

I noticed that in almost all open source projects, the main avenue of communication is the mailing list, or multiple mailing lists for big projects.

Every time I want to ask a question, I have to sign up, and it's usually not a very friendly or intelligent system. Memorable messages include: "Do not use a sensitive password because it will sometimes be mailed to you in cleartext" and "This is a reminder, sent out once a month, about your python.org mailing list memberships."

Then I have to mail my question in and wait for other people to answer, while in the meantime my inbox gets flooded with long threads about the project which are of no interest to me. After people have answered my question I still have to remain registered for some time, just in case someone will still have something insightful to say. All the while still getting unwanted messages. After that I have to interact again with the mailing list software to unsubscribe.

Why? Am I missing something here?

View on HN

Mailing lists are a way to collect high volume discussions on a topic in a way that can be quickly scanned and searched. I can route everything through GMail or other mail client, then use a single set of tools to manage my conversations and "tasks" that arise from those conversations. In my case, I've set up my mail clients to let me do most everything without taking my hands off the keyboard, just "shoveling" mail through via shortcuts. I also have mail filters that shunt different lists into different folders for searching later.

Web forums are fine for many people, but they drive me nuts. Too much clicking and too many pageloads, at least for the forums I have seen. This breaks my communication "flow" and leaves me feeling unwilling to spend any more time at the forum. It _is_ true that it is easier to drop in and ask a quick help question in a forum, as you point out, but that doesn't outweigh the fact I usually never want to visit the forum again. More fundamentally, my async store/forward communication happens in my email client, I don't like having to switch to other programs.

Now, don't get me wrong. Mailing list discussions are far from perfect. In particular it's easy for the entire discussion to be hijacked by a few hot topics. (Not necessarily trolls, although that works too.) This inflates volume and drives away people who would otherwise make great contributions. I've seen this repeatedly.

Really what I'd like is something that is to mailman what git was to version control -- a way for me to opt in to different "versions" of a mailing list potentially managed by different people but sharing the same core message flows. Something that would let me avoid needing to keep my own set of killfiles updated for each list all the time.

The Cypherpunks Distributed Remailer sort of tried this, but their focus wasn't on different filters. They never had that many nodes anyway. The other thing I've seen is people self-tagging posts with [<keyword>] in the title, but that is brittle and still requires me to create new filters on the fly.

Well I think the idea is that a mailing list, and by extension, the whole open source project, isn't a one way street. Most people who use open source software contribute in one way or another, whether it's through supporting new users, hacking on the source code, or offering helpful, informed suggestions about the project.

Not to be a Negative Nancy® here, but the problem is that "You're Doing It Wrong." If you're just wanting to get your questions answered without any sort of input to the project, I think perhaps you're missing the spirit.

That's all to say, it's fine for you to do that, but don't expect for it to be a frictionless process. The idea is that you don't do things that way. And most people who aren't (like me) really like the mailing list concept; it's easy, it integrates with something I already use all the time, and it exposes me to things and conversations that are going on really easily.

Theres more reasons to want to get support for something then your own selfish needs. Maybe you want to start hacking on the project but don't want to step on someone else's toes? Or how about if you're interested in learning about the project to see if it fits your needs (to avoid wasting your and other peoples time later on forcing a square peg into a round hole)?

And asking questions is giving input to the project. If a question keeps coming up, it should be sending you a signal that goes "Hey, this needs some work so that people don't have to spend so much time asking us how it works". I would argue that you're "missing the spirit" if you're arguing that its wrong to only want to ask questions.

but don't expect for it to be a frictionless process. The idea is that you don't do things that way.

Why shouldn't the process be frictionless? Why do you want to make it difficult for people to start using or get help using your software?

Theres more reasons to want to get support for something then your own selfish needs. Maybe you want to start hacking on the project but don't want to step on someone else's toes?

Then join the mailing list and participate in the discussion.

Why shouldn't the process be frictionless?

There's very little software that supports both fly-by questions on the web, and proper discussion via e-mail, so we optimize for ourselves, not users.

Web forums aren't frictionless. In fact, they're totally shit for the purpose of carrying on detailed conversations over the course of days without requiring constant attention from all the participants.

There's a reason why most forums will e-mail you when your watched threads are updated. Why not just have the e-mail be the thread?

Why do you want to make it difficult for people to start using or get help using your software?

Users don't pay the bills, and questions are usually incredibly repetitive and dull, no matter how easy-to-use or well documented the project is.

I'd rather the users that are unwilling to invest in the mailing list self-segregate by posting their one-offs to sites like StackOverflow.

It may just be me, but it seems like you're contradicting yourself with:

There's very little software that supports both fly-by questions on the web, and proper discussion via e-mail, so we optimize for ourselves, not users.

and

Users don't pay the bills, and questions are usually incredibly repetitive and dull, no matter how easy-to-use or well documented the project is.

While there's absolutely nothing wrong with coding for yourself, it doesn't usually pay the bills either. And (sadly), its rarer then it should be for someone to be able to print money for themselves by making things for themselves.

I reiterate: If different users are asking the same question repeatedly, perhaps its an indicator that there's something wrong with your software, not your users?

Oh, and, incase you don't know, not all software needs to expand until they can read mail. Multitasking has (sans the iPhone, if you want), been around for decades. People are more than capable of having one program open at a time. There's nothing wrong with this.

There's a reason why most forums will e-mail you when your watched threads are updated. Why not just have the e-mail be the thread?

Nowhere did I say that forums are better than email. They're both equally bad. My personal preference for this sort of thing is IRC. Of course, I'm biased, since one of the projects I spend the most time working on is an IRC client.

Also, forums are easier for end users to figure out then mailing lists. You could argue thats because mailing list software tends to suck, but, thats ignoring the issue.

I'd rather the users that are unwilling to invest in the mailing list self-segregate by posting their one-offs to sites like StackOverflow.

Not all open source projects are programming languages or libraries, which is SO's target area currently seems to be.

Feel free to treat "pay the bills" metaphorically. That was my intention.

I reiterate: If different users are asking the same question repeatedly, perhaps its an indicator that there's something wrong with your software, not your users?

Perhaps. But more often than not it's an inherent complexity in the problem space itself.

I don't follow your meaning ...

My personal preference for this sort of thing is IRC.

IRC is real-time and ephemeral, which makes it considerably more difficult to engage in conversations with contributors in other time zones, and refer back to design discussions that occurred 5-20 years ago.

Also, forums are easier for end users to figure out then mailing lists.

I thought we already established that project communication channels are primarily optimized for the project contributors who use them daily, not for once-off questions from end users.

Perhaps. But more often than not it's an inherent complexity in the problem space itself.

That isn't to say there is only ever one way of performing an action or that it is impossible to make it clearer how to perform a task. And if you make it difficult for users to give you their opinion, it will be that much harder to know what needs to be improved.

There's not very many projects that date back 20 years ago that also have the complexity that you're talking about. X11 and the Linux kernel are the only two I can think of off the top of my head. And I would hope that a 5-20 year old project that's continuously under development would have more documentation than posts from a mailing list about its design.

I thought we already established that project communication channels are primarily optimized for the project contributors who use them daily, not for once-off questions from end users.

I would be interested in seeing how many projects — open source in this case — have > 1 or 2 contributors.

I may be off base, but I'd guess that you're optimizing towards large, highly visible projects and not the small, norm of an open source project.

> That isn't to say there is only ever one way of performing an action or that it is impossible to make it clearer how to perform a task. And if you make it difficult for users to give you their opinion, it will be that much harder to know what needs to be improved.

Not to berate users, but sometimes when people run into an issue they want a solution yesterday. So they hop on to the mailinglist/irc and demand to have their problem fixed... Only for more informed people to point them at the FAQ which probably results for the first Google search on problem.

I'm not saying that the software could't have flaws. I'm saying that if people are armed with the answer to their issue, then they are more equipped to make suggestions at the future direction of the software. When they still don't have an answer they are still in, "FIX THIS NOW!" mode.

> IRC is real-time and ephemeral, which makes it considerably more difficult to engage in conversations with contributors in other time zones, and refer back to design discussions that occurred 5-20 years ago.

Lots of places log IRC and host the logs. That might not be your cup of tea, but it's far from hard. And I think that the reason that most design discussions (other than one-off decisions) still happen in email is that you can't really 'write up' something in IRC the way that you can compose an email.

> I thought we already established that project communication channels are primarily optimized for the project contributors who use them daily, not for once-off questions from end users.

Why not create a web forum/usenet/mailinglist gateway? Launchpad is semi-like this. I haven't pushed the bounds of it, but I can sign in through the bug page and post something. I can also subscribe to the bug and get an email for each post, which I can reply to (though it is limited in that there is no threading, but bug tracking is fairly topical) to post back to the bug page.

Lots of places log IRC and host the logs. That might not be your cup of tea, but it's far from hard.

In addition to privacy issues (many individuals don't approve of logging of what they perceive to be ephemeral 'hallway' conversations), there's the simple fact that these types of conversation don't lend themselves to well thought out messages -- as you noted with: "you can't really 'write up' something in IRC the way that you can compose an email."

Why not create a web forum/usenet/mailinglist gateway?

I'm sure nobody would object if you did -- Jive Forums (when the product still existed[1]) used to support both web-based and NNTP-based access.

[1] Jive Software was profitable[2] and engineering focused, and Jive Forums was a great product. They had both large and small customers, ranging from Amazon, Apple, and EA to numerous small businesses that required quality collaboration software -- jabber, forums, etc.

Then they took a massive, unnecessary $15 million investment from Sequoia Capital in 2007, hired a Chief Marketing Officer, completely fouled up their product line in "simplifying" it, and wound up having to raise another $12 million in series B funding in 2009 to keep the now-bloated machine running. They had to lay off a (reported) third of their staff in 2008 (but kept the 'CMO' position).

Waste of a good company with a solid set of collaboration products and a solid engineering team. Trimming down their product lines and raising their prices has left them out of reach of SMBs, and the "one size fits all" product completely ignores this straight-forward e-mail/forums/web problem space.

It also leaves a large market opportunity for someone to write a forums system that doesn't suck -- a surprising number of large corporations are still slaved to Jive Software's increasingly hefty licensing fees.

[2] http://www.jivesoftware.com/jivespace/community/jivetalks/bl...

Why did we raise the money?

Since the beginning, we have had a vision of open collaboration and how it can fundamentally transform the way a company works. Now the market is starting to agree with that vision and is seeing the benefits. We're proud of how we have grown this business over the last six years. We've been profitable since inception and have put good money in the bank. We have made our share of mistakes and missteps, but we haven't sacrificed our values and ultimately those mistakes made us stronger and smarter. This year we struck a mighty vein with Clearspace when we launched in February. Now the growth is in high gear and bringing on a funding partner is a step towards becoming the provider of choice in the market.

> I reiterate: If different users are asking the same question repeatedly, perhaps its an indicator that there's something wrong with your software, not your users?

Sometimes the issue is that people have different assumptions about how something should work. Sometimes there's not always an intuitive way to make something easy for every person on the planet to immediately grasp. But this is all a little too abstract without solid examples.

And here I thought the title of this submission was the opener to a geek-oriented comedy routine.

Email has been around for a while. People really like it. Older hackers (Torvalds) seem to in particular. I ignored it too. But I'm realizing the best conversations happen over it.

A lot of older internet users generally use lists (or private mail) for discussion. They also tend smarter. Like most scientist email and only a fraction blog. I don't think that's a bug. Probably better to extend email than force another medium.

Last time someone on HN asked about favorite lists it got nil response. Hope that was a fluke. But a good one is the Linux Kernel list. Intense. I don't code. But I'm learning a lot about how big projects get done.

The problem that I have with busy lists is that if you get behind in reading them it can look like a daunting task to see 100 conversations/threads (as in gmail) waiting to be read.

Site note: Filtering email lists in Gmail is really easy. There's a 'list:' modifier. So you can do 'list:python-list' if you've signed up to the mail python mailing list. Filtering them in procmail can be a PITA if you're trying to make a generic rule to filter list traffic into aptly-named folders, especially when traffic goes to two lists (e.g. python-announce and python-list).

"Do not use a sensitive password because it will sometimes be mailed to you in cleartext" and "This is a reminder, sent out once a month, about your python.org mailing list memberships."

Blame GNU Mailman, the absolute worst mailing list daemon ever written: http://www.jwz.org/doc/mailman.html

I don't think the problem is the mailing list style of discussion. Looks like it works fine, even for big projects like the linux kernel. What I find frustrating with mailing list is the archive:

* sometimes there is no archives

* sometimes the archive is unsearchable, only browsable

* for archives with search capability, sometimes the search results are not accurate enough

If end users can find answers through mailing list archive, they needn't post nor subscribe.

Mailing lists work exactly because of your complaint - they force you to stick around longer than you would with a webforum or IRC. If the list is good you'll say "yeah, I want to see what else is going on with this project." If you're purely an end-user it's much more frustrating, of course.

My main gripe about mailing lists is that I prefer using digest mode because it loads faster over webmail. But if I use it, it becomes non-trivial to reply to a particular discussion thread.

Really, that isn't a huge gripe, in the grand scheme of things.

As for "good lists" - pretty much any active project with a topic you're interested in will have some discussion.

Whatever webmail client you're using should get support for splitting digests. IIRC, there are scripts that do this if you use console readers like mutt. You should press your webmail provider/developer to add support for this. The digests will still be smaller since they lack a lot of the email headers that mailinglist traffic 'normally' send would have.

Oh also, if the project happens to have an IRC channel, those can be excellent places to ask these kinds of one-off questions. Depends on the project of course but instant messaging is pretty good for the quick help kind of thing.

In many cases you can send a message to the list without subscribing, but this isn't well documented.

In general I think the problem is just laziness; the people running the project optimize it for themselves (active participants) rather than their users. It doesn't help that the most popular software (Mailman/Pipermail) is mediocre.

> the people running the project optimize it for themselves

This is pretty reasonable, if you think about it: it's worth making life nice for the people donating hundreds or thousands of hours of their time to answer questions, as well as for the people who only plan to interact with the system for a couple of questions.

To the OP: Try the http://gmane.org/ NNTP/web mirrors of mailing lists, if you don't want to sign up for them. In particular, reading lists via NNTP from gmane in Thunderbird is pretty nice, with real threaded discussions going back as far as you want, and no need to sign up, etc. (screenshot: http://hcs.harvard.edu/~jrus/TextMate/gmane-thunderbird.png)

I really think the main problem is not the concept of mailing lists (and email/nntp) generally, but instead, as you suggest, the software that runs it. Google groups/mailman/etc. are all IMO pretty crappy from both maintainer and user perspectives, but they work well enough that no one's really putting effort into better mailing list software. Likewise, most mail clients are pretty crappy –either unconfigurable or with a difficult learning curve for configuration, with insufficiently easy-to-set-up and insufficiently flexible rule systems, etc. – when you get down to it. I should be able to absolutely instantly (<10 seconds) set up my email client to filter for only replies to my topics or mentions of my name in a particular mailing list (for example), but currently this is an involved process in every mail client I've ever used.

Of course, email is so so so much better for extended technical discussions than any web forum I've seen, that we all stick with it despite its flaws.

> I really think the main problem is not the concept of mailing lists (and email/nntp) generally, but instead, as you suggest, the software that runs it.

Making email better is undervalued. Love to see startups help extend it.

The problem is... How? Other than providing some sort of web front-end onto email (which is already done... Gmail.. Yahoo Mail.. etc), what can you do? You can try and use some sort of fancy email processing algorithms to do fancy things with email, but how does one make a business of that? Charge people to pass email through your system that sanitizes it and spits it out with better/cleaner headers for email clients to use?

I mean do more stuff through email, like Posterous and Xobni did. People are comfortable with the interface. I think Google Wave starts in email actually. But they extend it into XMPP.

You got me on the money equation. But I'm sure there's some way to make it.

Yes, thanks, I know gmane, it's a pretty good solution.

In general I think the problem is just laziness; the people running the project optimize it for themselves (active participants) rather than their users.

Unless your users are actually contributing money or time (most won't), I'd call that smart, not lazy.

This works and it doesn't. Some lists have disabled this, IIRC or enabled moderation (someone has to 'ok' the mail before it's allowed on the list) to stop spammers from trying to get their advertising onto mailing lists (presumably to reach the list audience as well as to get into search results through the list archives).

But yes, you can just send your email to (for example) python-list@python.org and then when people do a reply-to-all it will reply to you and the list... But what about a list-reply? (Gmail automatically replies to the list, not the sender) Does the mailing list software also forward mail from your thread to you, even if you're not subscribed?

"It doesn't help that the most popular software (Mailman/Pipermail) is mediocre."

Sounds like a business opportunity.

I did a survey on this for which to use with clients and came up with http://www.phplist.com/ which I've used a few times and found to be the best of breed OSS solution in this space. YMMV.

It was a while back, http://alicious.com/2008/best-online-mailing-list-management... , and I never really finished the public write up but put it online anyway ... isn't this the space that Google Wave is supposed to be filling now.

I doubt people would pay for better list software -- especially open source projects. People might pay for a better list service (ListHub?) though.

Yahoo! Groups was a mailinglist site that was swallowed by Yahoo! much in the same way that Google Groups was created by swallowing Usenet.

Never underestimate the power of inertia.

Never underestimate the propensity of the younger generation to blithely discount the previous.

Inertia is also what is keeping parts of the email system from being improved too -- though this could be a good thing if the improvements are 'improvements.'

Reply on news.ycombinator.com