A GUEST POST FROM DAVID DURANT, NOKIA, CHANGE MANAGER
Speaking to developers both within Nokia and other organisations about Symbian the response I usually receive is that people are very keen for the Foundation to succeed but, at least currently, they don’t feel that it has any impact on what they do. In short – where’s the love?
Developer communities lie at the heart of what any successful open source organisation is all about. It is not merely about communicating roadmaps or asking for feedback - it’s about building a community that would continue to exist if Symbian vanished tomorrow.
When discussing open source with people the first thing I always recommend is reading the bible of open source – Karl Fogel’s Producing Open Source Software (free download of entire book). It’s all good but I would specifically direct people towards the section on mailing lists (p48).
Fogel states “Mailing lists are usually the most active communications forum in the project and the ‘medium or record” as well as “Mailing lists are the bread and butter of project communications”.
Symbian supports a forum mechanism on its website but this requires either going to the site or subscribing using RSS (and then going back to the site). It would be very interesting to see the breakdown of visitors to the forums to see how many people are coming from established Symbian OS development companies.
Email, by contrast, is a simple very commonly used mechanism that allows anyone to interact via their own email client. Why am I focusing on mailing lists? Well, it’s all to do with reaching out to your community.
Currently Symbian is doing a good job at getting the word out to people at conventions, via blogs and through building commercial partnerships.
Unfortunately where I think the Foundation is weak today is in the area that should be its life-blood – connecting developers to each other.
Back when I was involved in helping build the Foundation internal infrastructure I suggested that a general Symbian interest mailing list could be created so people both within and outside the organisation could discuss the structure as it was built. I still feel this was a missed opportunity but by now, with everything Symbian does, this would probably have too much traffic so a different, more diverse, approach is needed.
My suggestion to Symbian would be to start by immediately creating email lists for each of the Packages. This not only gets the existing developers of that Package used to communicating via that mechanism but also allows interested outsiders a very quick method to begin talking to the current developers – without having to go through the Package Owner or involve Symbian in any way.
Communities can be quickly built around Packages but I would suggest that other mailing lists are created aimed at application developers using the various OS interfaces and runtimes (C++, Java, WRT, Python, etc). Other lists could be used to used to support commonly used Foundation technology such as Signing. Lastly mailing lists are an excellent mechanism to reach out to developers working in other open mobile platforms (Android, Maemo, etc).
A generic mailing list available to all the developers of these OS’s could start a great conversation on how they can best work together. Once these lists become active the communities they enable will very quickly become self supporting and they should be able to continue to work indefinitely without the need for any Symbian resources to maintain (although obviously it would be ideal of Symbian folks did take part).
It’s good to talk.
Agree with you 100% and this is what has made the Linux community successful. For example I was following dccp and netdev from http://vger.kernel.org/vger-lists.html when I was developing in the kernel and posting there.
Also another sign of health is when mercurial trees start popping up all over the place or we start hosting these similar to http://git.kernel.org
But David, there *are* package mailing lists. If they seem lifeless than maybe the Package Owners should be encouraging discussion? Certainly the ones I have talked to I’ve suceeded to get a response from though. Any comments about the ease of use, etc etc (they’re mailman so should be pretty standard) are a different matter.
Having said all that, this blog post isn’t totally redundant as your suggestion for more generic mailing lists is a good idea.
Creating a mailing list for each package and expecting a community to spring up is not going to happen.
A mailing list is an important characteristic of open source communities but it does not automatically beget them.
SF illustrates this problem perfectly by having ‘plenty’ of mailing lists: http://lists.symbian.org/mailman/listinfo, but most are unused (according to the archives).
I do agree that an active mailing list is a good open source health check, and mailing lists are the best way to hold technical discussions. However, I also suggest that you both scale the creation of mailing lists to how much interest activity there will be in the area.
Most importantly you need package owners who want to get involved in the community and can set the agenda for their area.
Dave, thanks first of all for the blog check. It’s very nice to see such an enthusiastic mention of the book.
Regarding mailing lists as the bread-and-butter: yes, I still think they are too, but even better would be to unify the web forums and the mailing lists to be the same entity (that is, make them present the same underlying information). A post I wrote a while ago, Thread Theory, talks about that in more detail.
It’s interesting that most sites still have this separation between their online forums and their mailing lists. I would have thought that the crucial distinction is between real-time conversations (i.e., chat rooms, whether in IRC or mibbit or whatever), and chosen-time conversations (i.e., most online discussion forums, email, etc). And yet the distinction communications tools keep making is between “email” and “everything else”. Google Groups is one of the few places that started to go down another road; I haven’t used it much, more out of happenstance than any particular aversion, and wonder how well the experiment is going there.
So I think there isn’t necessarily an exclusive choice to be made between mailing lists and online forums. Instead, it might be a matter of making sure the online forums are available as mailing lists, for those who want to use their mail client to participate. I suspect that group includes a disproportionate number of developers, who tend to focus on hyper-efficient tools and therefore shy away from using browser-based forum interfaces as a primary means of communication.
Best,
-Karl
Dave, totally agree with your observation that a thriving, kicking and vibrant mailing list has a snowball effect of gravitating contribution. However, each such list needs it’s champions who are keeping things together and answering questions etc.
Such bidirectional contribution is necessary to create value for developers. An excellent idea, as posted by tl, is for package owners to take that role…
@David, creating mailing lists by itself is not going to make a difference. There is a problem with momentum: we have already many mailing lists. One for all packages, one for each technology domain and many for individual packages.
The package owner and technology domain lists have some traffic today (slowly increasing) and package owners are automatically signed up to these lists. This means that everybody today can already get in touch with packages, should they wish to. And all package owners are very responsive when queries were raised. However that by itself is not enough.
Building communities depends to a large extent on getting noticed, evangelizing your technology and connecting to other companies. It also depends on your code being used by others: the two things are related by the way. Admittedly the foundation is playing a role here, but it cannot build communities for each package. Building communities is not easy and often there will be resistance from the managers of open source leaders. In some sense you are dealing with a chicken and egg problem in that it is necessary to invest time and effort in getting noticed and getting contributions, but for open source leaders to be allowed to spend the time, they would need to demonstrate that investing time will actially work for their technology (the generic argument does not tend to count).
The following article may also be of interest: http://developer.symbian.org/wiki/index.php/Building_a_community_for_a_package
@Karl: Interestingly I have been involved in discussions about exactly this issue: mailing lists vs. forums, which also in a sense is a discussion about push vs. pull of information. It seems there is no ideal technology out there today. Google groups seems to come close bridging the two, but not quite.
So maybe the problem is that we have too many lists? As the article writer mentioned a general symbian interest mailing list might be a better way to start with, and when discussions get too narrow or technical we branch it off.
The forums are quite good, I’ve seen several package owners comment within their own area, which is fantastic. We need more to do this, and I believe we just have to keep pushing and shoving and giving until the community can live on its own.
Perhaps it’s premature to expect heavy traffic on lists for various packages, when the code for these packages is not yet open source.
As packages move to EPL, and as it becomes easier for non-members to build and alter that code, there’s a greater likelihood for growth in discussions on the corresponding lists.
Excuse me David, but that does somewhat ignore the fact that there are plenty of people either working on packages or raising issues against the packages within Nokia.
Given that the Symbian support ecosystem has collapsed (to a default pay SOSCO for support who will be as clueless as the rest of us soon) it’s a pretty crap sort of Open Source.
Tony,
You’re right that it will be better when these discussions, which are currently happening on systems internal to Nokia, move (in general) onto the community forums and mailing lists. That will stimulate wider community involvement, and will help to spread the knowledge of all aspects of those packages.
This transfer will happen as package owners gain confidence in the stability and utility of the Symbian Foundation systems, and once they (and their managers) see early signs that more good than harm comes from this greater commitment to openness.
As will be discussed at SEE09, these early signs will include the growing pipeline of contributions (both major and minor).
I both agree and disagree with that last comment. The ideal situation to reach is one where engineers inside contributor companies always use the package mailing lists to discuss their work by default, so everyone can see what’s going on. However, until most of the packages are open source, the Nokia engineers are not going to be allowed (by their legal department) to do that. Whether working practices will really change after the shift to EPL is complete is a separate matter, but it is essential that they do for the platform to thrive.
Most of the platform packages that have been moved to EPL so far (and there aren’t many) are basically unusable on their own, so we don’t really have an open source project yet, we have one that will be open source in the next 6-9 months. It’s not pretty but it’s reality.
P.S. I think it’s wrong to classify email as a hyper-efficient tool (unless you mean one where the client doesn’t use many resources to run). There are more efficient tools from a time spent searching for information, reading and answering questions perspective – email and forums are about on a par in that regard. I’d tentatively suggest that “old school” open source developers are just more comfortable with email, and the new developers typcially serve their apprenticeship in established projects before starting their own, by which time the “old school” working practice is established and the pattern repeats. Email does have one significant advantage (for those that don’t use webmail at least) which is that it’s available offline. However, being stuck offline is becoming much less of an issue. Perhaps soon we’ll start seeing open source projects that aren’t “stuck with the communications technology of the 70’s” as one of my colleagues put it (taken slightly out of context).
By last comment, I meant the one from tl, before David answered.
> tl said:
I surely agree.
Talking about packages constituting the OS itself, package owners are mostly non-opensource guys, as far as I know (please highlight if I’m wrong). Those guys most probably are very good, but this doesn’t make them “open source dudes”.
It would be REALLY INTERESTING to see a table with all the packages, the name of the package owner, the company they work for (if applicable) and if they are package owners because of their company assignment (if applicable).
This will help a lot to understand the problem.
Out of the blue, I was thinking: why the Symbian Foundation requires a fee to participate and access the full code base?
Wouldn’t be more accessible if this fee was removed?
Wouldn’t it make more “interesting” if people could read the source code without having to pay a fee?
Just wondering.
Ivan, there’s already a full list of packages and their package owners on the developer website. As far as I know, every package owner is currently a Nokia employee and typically they are technical architects who were already working with the package in question (there are probably some post re-org exceptions).
The fee is only required to access SFL packages, all of the packages should move to EPL over the next 6-9 months, at which point everyone can see the code, fee or no fee. The reasons for the necessary evil of the SFL (mostly third party IP entangled with the platform and the need to check all the code to ensure it’s OK to open it up) have been discussed extensively in public already.
@Mark: As I expected. Well, probably Nokia needs to push a bit those guys to make them more “open”. I knew few of them and I’m sure they could easily write fantastic tech posts that guys like me out here would LOVE to read.
Not to mention the MLs.
The secondo point you answer too, was a bit of a “provocation”, to make someone repeat the main reason for which I believe the “community” didn’t take off.
EPLzation is done even faster than 6-9 months: I’m confident that more people will start hanging around here, once they can play with the source code of the operating system that runs the most of the smartphones.
Thanks for the replies David and Mark, and excuse my obvious frustration with the current state of affairs.
Even for a non-EPL package would it be possible to set up a list for those with the right licenses/agreements in place? When a package goes EPL it can become fully open. At least this forum can serve as a posting board for the contents of incremental releases for the package and discussion of integration issues or technology futures. Optimistically maybe this sort of mechanism would serve as a way of dipping a toe in the water of open source before being fully open.
p.s. Mark though email may not be ideal, it at least has been proven to work. Though their might be some subjectively better communication solution it certainly isn’t off the shelf web forums.
@Karl (and others):
“Regarding mailing lists as the bread-and-butter: yes, I still think they are too, but even better would be to unify the web forums and the mailing lists to be the same entity (that is, make them present the same underlying information).”
This is something I totally agree with – I personally find the web-based forums (both symbian.org and forum.nokia.com) quite cumbersome to use for daily updating, simply because they a) use up a lot more screen real-estate for the same amount of information, and b) come in a customized flavour for each site I visit, so all configurations I have made, say, to my Thunderbird or Outlook client, are worthless there. And using RSS feeds means a “media break” between the thread overview shown in my RSS reader and the actual representation in the web interface (sorting, highlighting of active threads etc.).
So while I believe web-based forums lower the barrier of entry, I think they are a burden of frequent contributors. As a data point, I have more or less given up following the Symbian Signed and C++ forums (which I used to do almost daily) after the loss of the NNTP interface that used to exist on symbian.com.
ciao marcus
Tony, I think we’re all frustrated at the lack of (externally visible) progress towards full EPL and open source working methods.
As has already been said, there are mailing lists where you can be sure the package owner is subscribed. We can and do create private mailing lists for specific working groups with access to SFL source that need to discuss it. You can request new ones in Bugzilla. We don’t have the IT infrastructure to automatically grant list membership to foundation members only on request, so this is all a bit labour intensive.
@marcus – thanks for the feedback around NNTP. Doing a quick google NNTP isn’t supported on vBulletin. I have captured this for any future changes we make to the forum software now.
Ian McDonald
Head of IT, Symbian Foundation
A group of us in Symbian have been thinking about this issue, prompted by the discovery that we did have mailing lists after all.
Mailing lists work for busy folk inside Nokia, because it is pushed to them (along with a torrent of email from inside Nokia). Having a newsreader interface to the same thing moves that from “push” to “pull”, and might suit people who want to read but don’t want to be answering all the time. Having the same information visible by a browser (and crucially by the website search engine) completes the picture.
Forum software is not pushed anywhere, so it’s possible to revise postings. This ability can’t easily be grafted into the mail/news world, but vBulletin doesn’t make much use of it. Good use of the ability to revise postings occurs on “Q & A engines”, where the answers are updated as the discussion goes on.
There are many higher priority activities going on, but I have a mailing list and an Ubuntu VM at my disposal, and have promised to try turning on the built-in mail/news gateway features of MailMan.
[...] Symbian Blog Official mouthpiece of the Symbian Foundation Skip to content AboutNewsEvents « Building Community [...]
I’m not sure the “push/pull” distinction is meaningful in this context.
Email and forums and news all have — or could have — the same basic UI: you go to your reader and all the unread stuff is there, presented in threads. The stuff you’ve already read, or marked as ignorable, is also there, just hidden, but still reachable and searchable (and marked with whatever metadata is appropriate, such as “I read this” or “I skipped this”).
So does push vs pull really come into it? I think the fact that some of these methods feel like “push” and some feel like “pull” is just an artificact of rather arbitrary UI conventions. For example, my mailreader and my newsreader are the same tool; if I still read news — I stopped years ago, when Eternal September began — news articles would show up in the same way emails do. I could organize things to give them a different priority, but that’s already necessary with email, since some emails are sent directly to you whereas others are via a mailing list, etc.
And if, say, the Symbian forum RSS feed were translated into an NNTP feed, then I could literally read and respond to those forums just as I do to email. (Actually, a better solution might be for my reader to be able to read the RSS feed directly, and perhaps it can; I just haven’t looked.)
Anyway, this is all getting a bit far afield from David’s original post. My point is mainly that thinking of all threaded, chosen-time information sources as being essentially the same thing can be a useful unifying principle in designing those systems (such as Symbian’s forums). The distinctions we have between these technologies are kind of a historical accident, the result of similar tools being developed in parallel, with the commonalities going unnoticed until much later.
@Mark: Hmm, email seems pretty hyper-efficient to me, depending on the email client of course. I can’t think of any more “more efficient tools from a time spent searching for information, reading and answering questions perspective”, but would welcome examples.
@William: When the forum software allows someone to update their post after it’s already been posted, that can still be problematic: replies to the post might be based on its old content, and stop making sense after the original is tweaked.
I guess this can be alleviated by marking any replies with some special notice that says “The (link) original post to which this is a reply has been changed since this reply was written. Click (here) to see the version of the original post that this reply was written for, or (here) to see the edit that was later made to it.”
Of course, that would gunk up the UI a bit.
“Communication tools are hard; let’s go shopping!”
I’d like to add my voice to the clamor for EPL/open source acceleration.
Open source is driven by passionate individuals who drive initiatives into their companies. The $1500 SFL roadblock cuts off individuals from participating and stimulating the Symbian community. In fact, I just got an angry email from a mobile industry veteran who had gotten excited by the forums about contributing to the QEMU effort, spent a day of effort, and then got stymied by SFL code.
Until it’s actually possible to build real open source community, what’s the point of asking over-worked Nokia package owners to add another communication mechanism to reach out to a community that cannot really collaborate with them?
Could we pick a week and ask the package owners to do nothing else but open source their package that week? Would it really take longer than that with a focused effort?
@Karl, re: email efficiency, it really depends on the specific project and developer use case. It would be completely impractical to follow all of the package mailing lists in a fully functioning Symbian Foundation platform and searching the archives for an answer to a question (or searching your own archived folders if you subscribe but just file them all) is a terrible way to find what you’re looking for – so you just ask a new question. Q&A engines like Stack Overflow are much more efficient for this kind of thing.
For genuine development discussions about a specific (reasonably sized) bit of software, I don’t see anything out there with a significant advantage over email.
We have all kinds of different developers in our community. Some fit the latter case but many (application developers and device creators particularly) are much more commonly in the former position – the trouble is they’d really benefit from answers from the guys who are in the latter position. How do we bring the two together with technology that suits both? It seems rather non-trivial. The various Q&A engines also have really useful collaborative features where question askers can mark an answer that worked for them, and other community members can vote for what they consider the best answers. This additional information helps people judge the quality of the information they’re getting, which is extremely useful in a large and predominantly anonymous community.
My personal view is that the mailing list works well as a tool for small to medium projects, but breaks down badly for very large ones, or those with big external APIs. For example, the qt-interest mailing list I consider completely unusable, and the Linux kernel mailing list is rather unapproachable for newbies. In both cases the volume of posting makes following them a massive time commitment, which tends to exclude a lot of people that might be able to make useful contributions in a more filterable environment.
I think there’s something really important underlying this debate. With open source, the community is the brand or image or identity or call it what you will. There isn’t a separation between this organisation and that community. The issues of building community around Symbian is partly that, right now, it is an organisation trying to work out how to build community; and it comes from a place where it was very clearly identified with organisations such as Nokia. We’re at a strange stage. Symbian presumably triumphs when it can say that “Symbian is the community”. We won’t be suggesting as David does that people wish Symbian well as if Symbian is an entity divorced from the mobile open source community doing well.The argument can go a stage further – if we all believe in open source, do we believe in an Android vs a Symbian open source community? Surely not – and surely that is one of the other strange things we are dealing with. A lot of us believe in open collaborative models but we are in a market that, at this stage in its development, pitches communities in opposition to each other. My guess is that ultimately we need to be advocates of open source and openness in mobile and treat Symbian as part of this sea-change in the industry.
@Haydn: I disagree. The “1 vs. 2″ scenario is the most obvious and natural (and frequent) because it actually makes sense. It’s Human Nature.
We all pick a side. And software is not a “neutral land”: we all have what we like and dislike. The same is for OSes.
And that’s just as phenomenon regulated by “singular human beings”.
Multiply this thing with the Monetary Interest. Android vs Symbian (for example) means Google vs Nokia(-SE-Samsung-(who else?)). I think it happens to often here to forget about this thing.
Of course, all those companies have, at the same time, multiple grounds in which their interests overlaps. But for the amount of investments that companies are pushing on those projects, do you think they are going to “take it easy”? No. That’s why a “1 vs 2″ scenario is the realistic one.
I also understand that the scenario gets even more complicated, but I think this is the bottom line.
While we spoke during the last weeks, and the Android blog was pretty silent, they released 1.6.
Don’t the Symbian Foundation thinks that this is taking a bit too long? I know I’m sounding very “unpleasant” asking this stuff, but while I read on a daily basis about phones of amazing characteristics from Nokia based on Symbian^x, I still see that the only way I can write a software is using either S60 SDK 3.x (or 5), and that Windows XP is the only OS on which I can do it.
In the last months I read 2/3 times different documents about proposals of Qt, Direct-UI and Orbit. All great. All good. Where is it?
What about a video of Direct-UI? I’m sure someone built a Flash-based sketch of the UI, right?
Orbit: wouldn’t be nice to have a wiki page with all the different planned widgets sketched-up and, possibly, in their preliminary look&feel.
Qt is the only place where A LOT of stuff is going on, with a GIT server that is literally COLLAPSING under the Trolltech guys pushing code.
In the meanwhile, Android moved from 1 to 1.6 SDKs + an NDK. Plus, they don’t have to worry about communities to build etc.: for them happened overnight. Mmmm…
My message here is: we all pick sides in life. And, as Symbian developer, I feel that my side is a bit weak and I don’t know where is it bringing me (us?).
@karlfogel
Sadly, “push/pull” is very relevant, because the Forum software we use can’t really do any of the things needed to make the distinction invisible.
You’d think that established Forum software would be able to indicate what you’ve read and what you haven’t, but that’s not true in practice with vBulletin. If it provided an NNTP feed, then you could handle it in your mail+news reader, but sadly it doesn’t.
I like your phrase “chosen time”, which seems to reflect part of what I feel is the push/pull difference. I try to resist it, but “You have new messages” is a dreadfully strong pull away from whatever I’m currently doing: “Drop everything – someone wants to communicate with you!”.
Hi all,
Maybe I arrive a bit to late to this conversation… but here I am never the less…
I would like to suggest that too much emphasis as usual is put on the technology behind this change – blogs, email ,forums, twitter… you name it , we have it all! So if finding a communication channel is a real problem, surely people would have found the way to communicate through the huge volume of channels hosted by Symbian.
It is more likely that the problem resides on the lack of urgency or perceived need to communicate.
I agree with @DW, probably EPLing will be the tipping point for Symbian. However, we need to do more to change the way that package owner engage with their communities, before we should start worrying about tinkering with additional support to our forums and email lists.