ideaTorrent playground set up

[dei]

Hey guys.

Just set up a little testing version of ideaTorrent this evening on my testing server. This system should bring a bit more order into design/feature questions and its proposed solutions. (Reading through the forums was a pain in some threads: SubjectA- SubjectB- A - A - trolling - C - C- B- offtopic-C-A-A...)

So here it is: http://ideas.goetz.sg

Apparently xenforo, being commercial software, doesn't wants to develop a bridge to any opensource software (Except Wordpress) and so you will (at least temporarily) have to re-register an account on this drupal-based ideaTorrent installation.

Feel free to play around with its functionality as we will flush the db-tables anyway before setting it up productively. If you wish to try out moderator or administrator functions, please leave me a PM and I'll promote you as soon as possible.

I'd plan to use it as feature request and design-discussion tool. Support-Questions and Art-Contributions (no upload in ideaTorrent) should be in the forums and ready-to-implement features and clear bugs should still be in github's issue-tracker.

What do you think? Could that be a useful tool to minimize reading-labour and moderation effort, and in the same time calling a vote for each solution?
Or complete crap? Why?

 

[Cervator]

I'm happy to be able to tinker with it, and understand it a fair bit better now. I've gotten ideas also for helping the forum here from it. It is a bit of a mixed bag itself tho

Pros:

• Definitely has more structure for ideas. Sandbox -> Validated ("Popular") -> Developing -> Implemented. Likey.
• Has an approval step where poorly formed ideas could be held back until solid enough to attract attention
• Could be used well for brainstorming different solutions to a particular high-level idea (bit of a different way of organizing)
• Available and in-use on SourceForge and also being used by Ubuntu with some 22k ideas in it
• Has a step specifically for finding duplicates that you can't skip (just a search) - likey
• Would focus topics much better

Cons:

• Project as it stands is abandoned That's a biggie, even if it might be picked up again later
• Can't currently integrate with our Xenforo (or our GitHub?) user system - also a big one
• Bits of the UI feel incomplete. Some pages don't update automatically after moving ideas, not always notifications for stuff that doesn't work. Bit confusing in places.
• The approach with multiple solutions to each idea might not be relevant some/most of the time? Often we just seem to have "Great idea! We do x!" and everybody goes "yay" and provides more related ideas in the thread. Often rather off-topic ideas as well
• The tool does seem to sit in an awkward spot between issue tracking "For sure and x is working on it" and forum "Hey what if?" + "Here are some in-depth implementation details" (and even the idea of a "design wiki" - tho that itself has also proven tricky)

So I'm not quite sure what to do about it. I do know that the forum is too much of a mess. I'm considering going back to a single "Incubator" dev forum for solid ideas instead of Design + Engine & World + Modding (where only contributors can post, but anybody rate and comment). I know we'll have content later to justify Modding, but right now we're just discussing at the same level across 4-5 different forums

I joked in the ideaTorrent about buying that fancy Xenforo ratings addon instead, tho I think we'll end up doing that anyway (it is just too snazzy in general). It would be a pretty minor part of an idea rating system, but if we did a copy-pasted header template in each "Incubator" thread and had the dev teams able to moderate the first post with updated status, GitHub ticket created, etc, it might work.

Neither solution is optimal. A more structured voting mechanism might indeed be nice, for stuff like the art style for instance (good example of "Idea" + "Option 1" + "Option 2" ...), but then we probably could just use forum polls. So would it give us enough benefit for the loss of convenience?

 

[dei]

I don't account ideaTorrent for the ideal solution but was fixed to it because there was no true alternative. Especially the thing with seperate user accounts and missing extensability/buggyness seems to be a system-killer.
Browsing through our own issue-tracker I had another idea during work today: We're using github to track features and bugs and it has most of the functionality we need for our idea-system too. It would be ideal if the transformation from an idea to an actual github-issue would be as seemless as possible.
So I thought of a a voting-plugin for github.

As github doesn't allow to install plugins I thought I'll try to build a wrapper using php-github-api (https://github.com/ornicar/php-github-api) which is supporting votes. And with this API it seems really simple to make.

So my vision (ordered by implementation priority):
1. In github we make a new category "ideas". (@Immortius: Could you help me with that?)
2. The user can login on ideas.movingblocks.net using his github account (via php-api)
3. The user sees a ordered (by votes) list of all issues within the category "ideas".
4. The user is able to add a new idea (issue with category "ideas"). (eventually separate rationale and solutions)
5. The user is able to vote an idea down or up with the press of a button. (is stored in a separate database)
6. Moderators are able to approve an idea for votes or reject it.
7. Moderators can convert an idea to an issue (ideas-ready-to-implement)
8. eventually: The User is able to assign a bounty for each idea/issue. (more toblerone for glasz )
9. more ideas?...

Is my implementation priority correct, or am I missing something?
I'll try to build a small prototype during weekend and present it here. IMO this infrastructure is more important than Terasology's coding-progress, else Immortius will die of a burnout syndrome having to tidy up all the forum...

 

[Cervator]

That actually sounds very interesting

Yeah, the lack of an alternative really sucks. If you're up for volunteering time to actually cobble one together, cool!

We can add an "Idea" label to GitHub issues just fine, that's easy. I wonder if that would end up swamping that system tho - I don't think we can exclude items with a specific label, at least not without explicitly including all other items.

But we could in theory using one of the other repos - we actually have a couple sitting around, like https://github.com/MovingBlocks/TeraMisc where I just activated issue tracking. That one is completely empty so you could actually use the categorization for categorization rather than trying to run two systems out of one issue tracker via label

That might also be a very cool thing for GitHub in general, there are bound to be users out there who'd love that. You could even commit the code to that very repo, as that is exactly what it is for (misc extras). Let me know what your GitHub user is (not sure if I've got you on there yet, I don't see a "dei") and I'll give you push rights to TeraMisc, that should allow you to tinker with the issue tracker (I think)

I'm still curious what we could do within Xenforo too, maybe some combination would be ideal. And I agree that this sort of infrastructure is very important. We just have to find a balance and I'll be happy to front on the forum and GitHub angle so Immortius can keep coding

 

[woodspeople]

Here's a test case. I/we have been planning/hoping to do something helpful with the minion design questions thing, to make it more helpful towards API requirements (assuming of course that WOULD be useful).

My/our thought was to:
- take the original list of questions and dump it into a spreadsheet
- use the spreadsheet to cluster the questions into groups (like 5 groups or so)
- for each cluster, create a Xenforo forum post with poll (bottleneck: lots of copy and paste)
- ask people to "vote" for some number of items in each poll as being most important to consider (no way to enforce that in Xenforo poll apparently, so just need to ask people to count their own answers...)
- ask people to add more items to lists in comments to each thread (then somehow people can vote for these by ...)
- after some time, compile all votes and assemble list of questions to be considered in more detail (say 20 highest ranking or something)
- for each question, create Xenforo forum post with poll and open-text questions (to be answered in comments), things like "think of a mod that might answer this question, describe what sort of infrastructure support would make that mod easier to create"
- ask people to fill those in
- take results of that and make document outlining "support infrastructure to best support proliferation of cool minion mods"

This is an awkward fit for the Xenforo forums. How would something like this work in GitHub or ideaTorrent? The goal is to keep the process quick and easy, yet productive.

 

[Cervator]

I think that might be too much text / threads / polls anyway. Clustering the questions logically by topic and such would be what would really be a huge help, then post a thread for more brainstorming on that area (more questions to add?). We can hook up the Rating plugin (it isn't free but I can sink the money for it) and users can then rate individual posts to get a feel for which additional bits are the most interesting (we can do custom ratings like +1, +2, +3...). Could do the top poll there too. Calculating exact actual details (votes, etc) might be too much effort manually, more of a "feel" thing for "Oh yeah, this looks like it has lots more positive ratings"

I realize the design process sort of relies from separating the questions from the answers, pushing those to later, but I wonder if we can resist trying to answer stuff up front with potential answers? That's sort of like what ideaTorrent seems apt for: "here's a topic, now submit questions and rate them!"

Making a thread for each actual question then sounds like what would scare me, but I suppose the point is (again like ideaTorrent) that you only do a few actually popular questions at a time, so maybe I'm just too skittish. Certainly not against trying, tho I think a lot of the coder-types we're likely to hit would want to start coding sooner

The Incubator forum I'm thinking about would probably be similar. One main topic per thread, some template stuff at the top (relevant links, phase, anybody who has volunteered to work on it, rating options, etc), then discussion, design, and implementation as they develop naturally in the thread.

(Disclaimer: Am frazzled at work, have not gone over this a few times to really think about it as I usually do, so take the details with a grain/teaspoon/cup/truckload of salt)

 

[dei]

Cervator: My Github username is cephos . Because dei was too short...
Perhaps a good idea to use an own issue-tracker keeping things tidy. What would be the disadvantage is, that we would have to recreate the issues for development. (What would have been the case anyway, ok.)

 

[woodspeople]

Yeah, if I was doing this sort of brainstorming activity in a room with people, we'd just use sticky notes and the whole thing would flow quickly and naturally. Trying to translate collective design processes into the artificial, constrained flow of typing and reading texts makes everything awkward. The question of how to facilitate group sensemaking online is one people have been trying to address for decades, and it has no wonderful answer yet as far as I know. One of the best options is to use a "whiteboard" people collaboratively fill up with design ideas. Some people like to use "mind mapping" for design (google that). I like it some, but don't like how hierarchical it has to be (no way to just group stuff together) and also it tends to lead people to spend a lot of time "prettying up" designs. Too little like the back of a napkin, which is where all designs should be. Which is why I like whiteboards, and wikis are text whiteboards (at least the easy-to-use ones). There are some collab whiteboards online, not sure how free/useful any of them are. I have had some good success in dev teams using collaboratively edited documents (spreadsheets, text docs, diagrams).

I don't think anybody could resist trying to answer questions up front. If you read my questions carefully you can see that I embedded some answers in them (like, for "how will the player oversee minions far away" I put in the "minioncam" thing which is not a question but an answer). I meant to remove all those in the spreadsheet version of the questions. Maybe - this is an idea - instead of asking people to vote on questions, ask them to pick some questions to answer. If a question seems like a fun one to answer, that's a vote. Might be a better way to gauge importance.

Just looked at ideaTorrent a little. Two problems with it. One, elements can only be ideas. This would be great for feature requests, but for design you need a more flexible way to move around lots of other kinds of elements. Hence the whiteboard/wiki stuff. Two, the popularity stuff is not right for design. For adding new features sure, but for infrastructure, it just gets in the way. I could imagine ideaTorrent being great for user ideas though.

 

[Cervator]

Yeah "popular ideas" really is a poor label, thus why I sort of retconned it to "Validated ideas" in my little write-up. It looks really promising but incomplete. But it would be its own whole project to take it from where it is now and complete it. And probably likewise on grafting a poor man's version onto GitHub. I'm not going to turn down an volunteered attempt tho, especially if it integrates nicely with GitHub and/or Xenforo.

GitHub issues aren't great for lengthy discussion, that's more up Xenforo's alley. The two combined with some of the rating stuff from the Xenforo style tho - madness.

More thoughts stuck in my head but I want to go home

cephos - yeah, I've had to ask for that more than once, haven't I. Added you to the contributor group now so I can remember, woodspeople's account too. Added both of you with push rights on TeraMisc in case of tinkering - I think that allows you to do stuff to the issue tracker

(There's probably some way to automate the process to "promote" an issue from one spot to another, or what not..)

 

[Immortius]

1. In github we make a new category "ideas". (@Immortius: Could you help me with that?)

Finally found where to do this, so done for the main repository (tag is "Idea", to follow the existing style).

else Immortius will die of a burnout syndrome having to tidy up all the forum...

I don't actually have any access to tidy up forums at the moment, so no risk of that. But I imagine Cervator must be a little frazzled by now.

 

[Cervator]

I don't actually have any access to tidy up forums at the moment, so no risk of that. But I imagine Cervator must be a little frazzled by now.

You might not have realized you have said access...

(Purple user text = forum admin)

 

[Immortius]

Maybe I'm just missing how to do it then. I'm not seeing any admin controls appearing. I'll investigate.

Edit: Ah, the super moderator thing you mentioned earlier.

 

[dei]

I meant Cervator in my last post, not you Immortius. Somehow mix you up all the time....
We'll do the idea-thing in the TeraMisc Project on Github, so I don't need the idea-tag in our main project. I already have the needed access rights to create labels etc.

 

[Cervator]

Okay, more thoughts, a new prototype forum, a crazy design idea involving Xenforo, GitHub, and some dei magic oh and making him a site admin to help tinker with said crazy ideas

I put an "Incubator" forum under the Development heading, more or less a fancy redo of the old Design & Implementation forum, meant to replace Design, World, and Modding till we can justify separate forums (not yet - but we still need to structure better) - intro post here

I used some material from earlier discussions including the idea from woodspeople about tagging threads with some detail of where they're at / what they're about. I made a header meant to go in the top of each thread, akin to the Contributor Intro forum, and one of the fields is for having a "Curator" who is the main person working on whatever it is. There are also some other bits that might have some magic potential (automatically find and display some stuff)

At the same time I wanted to have a way to integrate that better with the ideaTorrent approach and Woodspeople's "questions" design style. Mixing in Dei's thing to add some sort of voting on GitHub issues gave me the idea to label issues in the TeraMisc repo with the name of each feature with a thread in the Incubator (tho likely only some will get the full question-answer treatment) - then we can pull in the related issues (or "questions") easily to the feature thread itself, and maybe find a way to vote on them there via some gadget Dei puts together - which could also do something separately a la ideaTorrent. Some or maybe even all or it could be setup within Xenforo - it has addons for authentication via Facebook, Twitter, etc (or associating your account there with a forum account), so I bet GitHub would be doable too. Maybe for these mini-issues we could have a mini-form in the forum to submit simple "questions" to add to the design phase. Key is making it simple and easy to use, otherwise it won't be used.

Later that leads naturally to "mods" (in both senses of the word, both central core content shipped with the game and custom content supplied by modders) also being organized that way with a link to a category of issues on GitHub and a "home" thread in the incubator (maybe an index sticky when it gets big?). This sounds similar to the "Design Wiki" concept but in a format we can more easily handle in the forum vs the pretty light-weight wiki. Then you could also have bugs, enhancements, etc, that can also be applied to a feature label (at which point "question" might also be needed to distinguish the design system entries from more normal issues).

Some mods will be done entirely externally by some users, but (unlike MC?) I'd like to see a large number of useful mods maintained closer to core than just off somewhere in the aether (see the Jenkins project for an example). The Terasology repo itself could stay focused on the engine and project itself, giving reason to keep two separate issue trackers (and eventually separate to both an Engine Incubator and a Modding Incubator).

All the while keeping in mind the "shortcut" factor Woodspeople also has highlighted - we won't have that fancy integration for a while, unless Dei keeps multiple crates of 5-hour energy handy, so in the meantime we can start out with something we can expand on later - which could be the Incubator forum for now, with manual headers, which would be one of the primary responsibilities of the Design Team to help maintain and organize

Haven't quite got all the exact details in my head down here, but it is 1 am and myself plus wifey have dental appointments in 9 hours - yes, on a Saturday

Thoughts?

 

[Cervator]

Few more summary notes from the meeting:

1) dei is tinkering with an idea/voting app based on GitHub issues, exact design still a bit up in the air (Xenforo admin powers going Dei's way, but might need more tweaking by begla - not getting menu items or something?): http://andre.goetz.sg/ideaTorrent.movingblocks.net/gitvote/idealist.php
2) woodspeople is suggesting tinkering some with Google Docs or other collaborative brainstorming apps to do the design questions / tiny snippets you'll want to glob together a bunch of to start shaping a concept / a grouping of things that can make a feature. This might be what Dei could be putting together some sort of mash-up for. I think we'd all be happy to add tons more questions to a topic when we have a format to do it in
3) Cervator is working on the Incubator forum angle that's sort of for when you have a glob of small ideas you're ready to incubate into cheese. Or rather a concept that can proceed through the four stages of inception, design, implementation, and maintenance

We have more to discuss on this I'm sure, like forum descriptions and such. There was something orthogonal about the Incubator to look at too...

 

[woodspeople]

Looked at online collab/brainstorming tools. Lots, and some cool ones, but all cost in the range of $5-20 per month for private groups with saving, etc. Best open source alternative we see is FreeMind http://freemind.sourceforge.net/wiki/index.php/Main_Page It works off-line BUT mind maps can be viewed online (see example here http://freemind.sourceforge.net/PublicMaps.html). We could put some of the minion questions into it to see how it works. Just tried with a few - you can move things around within groups, so clustering can happen. Ideal thing is online collab with real time interaction, but succession games can work as well ...

 

[Cervator]

Looks cool to me

Happy to give it a spin if you put together an initial base map to then play succession games with?

 

[Cervator]

Added some updates to the Incubator forum - moved some threads around, renamed some stuff, and lined up some good examples:

• Creature Definitions - woodspeople "questions" thread originally split out from miinions and an example of Engine scope, as it relates to how we define the very fundamental ways to describe and make creatures functional
• Miinions - overdhose concept, example of Core Content scope as it outright provides actual content in-game. Currently is flavored a little, might become a core system with more flavor in an actual mod (see next)
• Manus Lenoni's Purple Hand of Doom - example of Mod scope, as it would be optional and full of DK flavor as per desire from overdhose, including the infamous cartoony slapping by a big disembodied purple hand

Those three also go in sequence - Creature Definitions provide the very basic "what are creatures?" design, from which one category is implemented via the Miinions concept, which could be further flavored later by the MLPHoD mod

So one could spawn the second which could spawn the third. Yet each can go all the way to implementation complete independently, and also provide a feature-specific home for useful links, docs, auto-listed GitHub issues later, etc.