search

euruko2013 / committee

EuRuKo 2013 organising committee repo
6 stars 0 forks source link

Euruko 2014 selection process #4

Closed damphyr closed 11 years ago

damphyr commented 12 years ago

Several comments on the 2012 Euruko said that the voting does not really scale with the conference. Brainstorm about ways to make voting "fair".

petros commented 12 years ago

Are you referring to the voting procedure that elects next year's organizing city?

damphyr commented 12 years ago

Yep. I had an idea on using tokens and a web app to do it, but as @pagojo mentioned the other day, that will take the whole party out of the election process. So we might want to come up with something that can be staged live as well.

petros commented 12 years ago

Hmm, so we want fair + fun. Let me sleep on it.

pagojo commented 12 years ago

Assignment:

During a conference, delegates are called to vote between 4 choices by show of hands. This is tedious for the organisers that have to count and recount votes as well as for the delegates that have to keep their hands raised. What we want to do is to allow them to vote by show of a coloured piece of papers. There should be 4 such colours.

When the delegates vote they will have to show their choice by raising their hands holding this coloured paper. Subsequently the system should recognise how many votes each colour's category receives.

This is a problem which involves machine learning, pattern recognition, some hardware interfacing and some image processing.

The things to look at are the OpenCV library, K-NN and SVM classifiers and the SVMLib. This problem should ideally be implemented either on Ruby on the desktop, or on a mobile device running Android or iOS (on iOS one can also use RubyMotion whereas on Android one can use Ruboto). Training of the classifiers will be offline and probably may involve some scripting or command line tools.

petros commented 12 years ago

@pagojo This excites me.

damphyr commented 12 years ago

Down boy! :P I like the idea, but having done something like this I expect several difficulties not directly related to the software (lighting, camera positioning, movement blur etc.) Easiest it would be with cards that have distinctive patterns easily discernible from one another and with high contrast and vivid background colors.

petros commented 12 years ago

To be honest, I don't have any idea how reliable something like this would/could be. The last thing we wanna experience is this system failing and not having a plan b.

pagojo commented 12 years ago

Plan B is always 4 people counting <= 500 raised hands....old tech but it works!

I was thinking that such an assignment si ideal for an MSc student btw so I have already contacted some academics :-)

Anyway it is just an idea...

damphyr commented 12 years ago

OK, I was thinking about the best way to get a photo of 400+ people holding up cards so that we have minimum overlap. The obvious solution is an overhead camera. Then I read this http://www.wired.com/dangerroom/2012/06/ff_drones/ and since I did a bit of research for a TU project proposal we did at work I've already looked at how much these cost and what can be done with them. Now imagine a quad drone flying overhead and snapping a picture that gets analysed to produce the voting results. That would be cool! :D

apantsiop commented 12 years ago

There is one other factor that we should consider. The geography (or geometry) of the room. Most conference rooms are semicircular so a camera that could take such a picture would look like this http://en.wikipedia.org/wiki/Fisheye_lens . Also the solution has to be something reliable, as well. The drone idea is way to cool, but imagine the awkward moment of a drone accident/malfunction :)

pagojo commented 12 years ago

Shall we go really wild and make this into a competition?

The team which creates the software that works gets to host the next EuRuKo and if more than one team does, then we use their software against them in a crowded room, or something like that.... :o)

of course we need a backup plan in case none does....

are we allowed to do this? Are there any specific rules as to how the next nomination should happen?

pagojo commented 11 years ago

Seriously now what do you think?

pagojo commented 11 years ago

OK, who will pick up this hot potato? @amiridis are you up to it (please:-)?

apantsiop commented 11 years ago

Shouldn't we brainstorm before assigning/implementing? On Nov 22, 2012 11:18 PM, "pagojo" notifications@github.com wrote:

OK, who will pick up this hot potato? @amiridishttps://github.com/amiridisare you up to it (please:-)?

— Reply to this email directly or view it on GitHubhttps://github.com/euruko2013/committee/issues/4#issuecomment-10645228.

vvatikiotis commented 11 years ago

+1 @apantsiop

pagojo commented 11 years ago

@apantsiop yes 100%!

At the same time since it has been really dormant this one (5 months since I posted the last question), I thought we should apply "gentle pressure" for someone to own it. After all it was @amiridis who said "This excites me." :-)

vvatikiotis commented 11 years ago

@pagojo Does this allow for other ideas or are we set on doing this particular one?

pagojo commented 11 years ago

@vvatikiotis "particular one"? I don't think there is a particular one, I just suggested @amiridis or someone else picks this one up so that he coordinates the investigation/brainstorming for our options etc etc.

vvatikiotis commented 11 years ago

particular refers to this https://github.com/euruko2013/committee/issues/4#issuecomment-6448290 . I'm all for someone investigating possible alternatives in general to the voting procedure.

vvatikiotis commented 11 years ago

tbh I kinda like the the-loudest-wins method. It's quick and very vibrant and I guess exciting. It sounds like an energetic and festive way to close an event.

OTOH, whatever rocks your boat...

pagojo commented 11 years ago

@vatikiotis sure, even this one needs investigation as to how it can be implemented though. For example I have no idea how it can be done

vvatikiotis commented 11 years ago

@pagojo fear not, @damphyr to the rescue!

petros commented 11 years ago

:ok: challenge accepted. Can everyone please dump their brain in here? Remember people, this is brainstorming. We don't need to criticize each other's brain farts yet. Just say whatever comes in your mind and we'll take it from there. :smiley_cat:

I'll dump mine tomorrow because right now my body runs on reserve battery.

nolamesa commented 11 years ago

Ok since we are not criticizing anyone here's an idea: Similarly to the shouting concept we could ask people to do a jump (maybe in a wave style) and count the vibrations. Not sure how hard/unfair this would be, but it could be fun...

nikosd commented 11 years ago

Trying to tear down the venue? :smiley:

pagojo commented 11 years ago

I was thinking that we may wanna play it safe here. One of the things we could do is to have the ticketing service issue a unique (ticket?) code during ticket purchase. When the time comes to vote for the next EuRuKo each delegate could enter this code and the email with which they registered into a form and vote that way through a website (during a 10 minute slot on the last day). Boring perhaps, but worth it as plan B?

vvatikiotis commented 11 years ago

Closing

nikosd commented 11 years ago

@vvatikiotis why did you close this issue? It's not completed, right?

vvatikiotis commented 11 years ago

You are right. I had in mind the speaker selection process. Reopening.

pagojo commented 11 years ago

We have volunteers working on plan a) a computer vision based voting system (Panos Adam) and b) an extremely basic forms based voting system which will use the Eventora data, like the one I suggested in this thread (Giorgos Batistatos).

both have reported progress and will have them for demo in the next two weeks. For plan A we discussed that we need to vist Badminton Theatre and take some snaps as well.

nolamesa commented 11 years ago

As discussed in the previous meeting there is a chance that we might fail with our Plan A option. So I am thinking to try to implement last year's option as an alternative. I have already requested from @fotos to ask Josh what they did last year so that we can get any existing code/specifications. Meanwhile, I was thinking to tell Basistatos if he can do this as well. A good idea would be to ask help from the guys over at hackerspace.

Shall I move forward with this?

pagojo commented 11 years ago

George Batistatos has things to finish up with the backup plan anyway, so we need to find someone else. Panos Adam is working on the image detection options as you know.

We should discuss this at our meetup

fotos commented 11 years ago

This is what I proposed on the last meeting and discussed briefly:

Let's show 'em how it was done by our ancestors...

The standard format was that of speakers making speeches for and against a position followed by a general vote (usually by show of hands) of yes or no. [...] Voting was by simple majority. [...] At the end of the session, each voter tossed one of these into a large clay jar which was afterwards cracked open for the counting of the ballots. Ostracism required the voters to scratch names onto pieces of broken pottery (ὄστρακα, ostraka), though this did not occur within the assembly as such."

So 2 ideas stem from this:

Idea 1:

  1. A narrator from each candidate city speaks (only a single rhetoric is allowed by the best narrator, no images, no powerpoint, no video and no sound unless they play the instruments themselves) for 5 minutes. It should be imaginative, descriptive, playful, persuasive.
  2. We vote by raising hands for each candidate city after the end of the speeches for the candidate city to reach the next ground Shall 'Paris' move on to the next round 'YES' or 'NO'?, until 2 cities remain to battle it out. After each round the remaining cities get a 5 minute speech if they want.

Idea 2:

  1. Same
  2. We vote using "mobile" ostraka. :) Codigia creates a hidden "Vote" menu (named "Boule" or "Agora" or something) that appears after 17:00 on the 2nd day. It pulls data from "candidates.json" (for the cities and perhaps a description). It resembles an ostrakon (see the image) in a old-school Hellenic font where you can strikethrough the names you don't want and leave the one you want. Or the opposite.
    • Pros / cons for idea 1:
  3. Low tech, simple, effective and subjective like the good 'ol times! Less work necessary! :smile:
  4. Not accurate (this is both a pro and a cons depending on your PoV - the decibelometer was not accurate BTW)
  5. Fun
  6. Old-school
  7. Shows a bit about ancient democracy
  8. yada nada
    • Pros / cons for idea 2:
  9. In accord with the digital times we live in
  10. Accurate (this is both a pro ans a cons depending on your PoV - the decibelometer was not accurate BTW)
  11. Fun
  12. Hipster
  13. Shows a bit about how we have progressed since ancient democracy
  14. Needs WiFi / web access
  15. yada nada

In order for this to work we have to:

  1. Tweet it in time (midweek). Teams interested in getting next year's nomination should prepare an orator and a speech (we mean it). We let them know, in advance (2 weeks before or even now) so they don't lose time preparing videos and presentations and stuff but speeches instead!
  2. Lay out the rules
  3. Find a chlamys, the backdrop and a(n image of a) statue
  4. Prepare the app(s) based on the outcome of this vote

Setup: As a backdrop on the project we display a nice picture of Pnyx, an ostrakon and the statue of an orator (Demosthenes, Gorgias, Plato, or another from the list of 10 canon orators).

I, personally, trust in Codigia to produce an aesthetically yet functional and user-friendly application. We can also provide a simple web app, and use the backend (continuing on the work of Batistatos) for POSTing data from the mobile app. We can also present live stats of something. But this needs dedication and time, and I don't know if we have them available. Also it's risky.

Please vote on which idea you like / prefer (but keep in mind the work necessary to do it, if you are not going to be involved in its making). Or come up with a better one. The vote is until this Monday (24 hours).

Personally I am willing to coordinate Codigia (like I did before with Diamanto), Batistatos (to create the web app / backend, review the code and stylize it) and to wear a chlamys or chiton for the happening. This is Athenssssss! :stuck_out_tongue:

fotos commented 11 years ago

Ohhhh and my vote goes to Idea 1 which at this point prefer due to its low tech. :sunglasses:

vvatikiotis commented 11 years ago

About the low tech

About the setting (narrator etc). I like it.

About the voting voting mobile app. I like it.

If we are going with electronic voting: It's not going to be another app but integrated with the Euruko app codigia already is preparing. John has already the task of the voting app. The voting backend is almost ready but we can enhance it. Any work that needs to be done is another section in the Euruko app + gluing the mobile app with the backend, which is trivial (POST). John can coordinate the whole thing since he's already on it and knows Stefanos personally.

We'll need a vote MC wearing chiton/chlamyda :D If we go wuth us presenting the event then the vote MC can be the event MC. Or the duet. Whatever we decide.

PS. Why are we trying to reinvent everything? Frankly I'm pissed off.

pagojo commented 11 years ago

For the electronic voting: a) JSON API is done in the voting server, by George Batistatos,will test later today b) Voting can be from the app and from the web (so that we can cover all bases) c) Voting and access to the voting server will be restricted to the netblock inside the venue during the days of the conf :-)

So I see no reason for discussing option 1) :o)

apantsiop commented 11 years ago

@vvatikiotis Fotos said we will not be counting, we will base our decision on a quick optical assessment. Which makes it quick BUT even more inaccurate. I know the decibelometer is not accurate either but it has a degree of accuracy/fairness. The proposed way could be heavily scrutinized afterwards. I go for idea no 2. And the chiton should be worn by the contestants not us.

nikosd commented 11 years ago

I prefer low-tech than high-tech for the following reasons:

  1. Everyone can participate - no device required
  2. We don't depend on connectivity/networking/etc. It's 100% fail safe unless the lights turn off :D
  3. After so much technology a time out sounds nice in my mind

Having said that I'm totally ok with whatever.

damphyr commented 11 years ago

Hehe, how about simplifying the voting: We use scales! We put up as many jars as there are candidates. Everyone gets a marker (call it an ostrako) and queues up and casts it in the jar. Whichever comes up heavier wins. This is low tech and rapid feedback and it has a low error margin.

fotos commented 11 years ago

Just to clarify the thing about the decibelometer (hopefully once and for all): it is not accurate by any meaningful interpretation of accuracy. If I can scream even a bit louder than you can, then my vote has a totally different weight in than yours. This makes it inaccurate. And I'm not even talking about technical limitations or error margins. The most important thing in democracy is that every vote is casted equal, regardless if you are rich, poor, black, white, tall, short, in the government, a peasant etc. etc. If we want accuracy we either go electronic or implement low-tech ideas like the one proposed by @damphyr (you 'll also find it in the description I posted above about the clay jar). I kinda rejected this idea initially because I thought you wanted the process to be fast... anyway it's a nice alternative.

pagojo commented 11 years ago

Chaps the electronic voting is almost a done deal in terms of tech. Few remaining masterstrokes from the marvellous Codigia duo (Giorgos+Stefanos) and we're good to go. George Batsistatos has created the JSON APIs (which we will modify a bit asap) this weekend already.

vvatikiotis commented 11 years ago

Super john. Update your status update while you are at it.

On Monday, June 3, 2013, pagojo wrote:

Chaps the electronic voting is almost a done deal in terms of tech. Few remaining masterstrokes from the marvellous Codigia duo (Giorgos+Stefanos) and we're good to go. George Batsistatos has created the JSON APIs (which we will modify a bit asap) this weekend already.

— Reply to this email directly or view it on GitHubhttps://github.com/euruko2013/committee/issues/4#issuecomment-18868959 .

nolamesa commented 11 years ago

I think the low-tech solution, as presented, does not "converge", if we don't get exact counts. Last year's voting was very close. In a similar scenario we would have to do this for hours. So in this case we would need to count every hand (not so tragic, we could seperate the audience in blocks and count one at a time). Another issue that I can see with the solution, as presented, is what is the "optical threshold" above which a city moves on? Is it half the audience and how do you compare the optical amount of votes for the first city with the one for the last. Maybe I missed something here.

The high-tech is a bit less fun but I am not against it. Two important issues: it has a high risk of failing (it's gonna be untested production code) and people need a device for it. The bad thing about the latter is that people might not have battery at their devices (end of event), they will have to skim through their emails to find their ticket id (this makes it difficult to vote from someone else's pc) and we could even get some connectivity issues (wifi? server not responding due to load?). As a result many people would get excluded from the voting. Btw we will also need a website version of it in order to be proper.

Overall I would go for the last year's solution which is elegant, easy and fun (though not accurate as @fotos said). Besides it is also one of the ancient ways of voting. If we don't go for this I am for the low-tech with counting (I can't really see how it works without it).

In other matters I disagree a bit about not having ppt for every presentation. We are geeks and not politicians :wink: plus the images are great for the audience's imagination as well. Remember it would have been hard last year without an image of a greek island :smile:

pagojo commented 11 years ago

The high-tech is a bit less fun but I am not against it. Two important issues: it has a high risk of failing (it's gonna be untested production code) and people need a device for it. The bad thing about the latter is that people might not have battery at their devices (end of event), they will have to skim through their emails to find their ticket id (this makes it difficult to vote from someone else's pc) and we could even get some connectivity issues (wifi? server not responding due to load?). As a result many people would get excluded from the voting. Btw we will also need a website version of it in order to be proper.

Let me address some of these reasonable points and hopefully convince you :-)

  1. We will tell everyone the day before to note their a) email and b) ticket number so that they may vote
  2. It is probably a safe bet that delegates of this event will have at least one device that can connect to the web. We can also tell them to have them juiced up. If we think that many people won't carry devices at all then why are we bothering with outlets and WiFi in the first place?
  3. If the WiFi is not to work then we will know in advance, on Friday morning to be exact. In which case there are other panics to consider and at which point we will have to revert to manual counting.
  4. Any software or server may fail. Granted this is true, however we keep using software. If it fails then we can always count hands. But it won't. I trust @nikosd review of the code and the known gem which George used. I'll also do my best with G. Batistatos to make sure the server is up to it. If this voting software can fail, so can the decibel meter scheme and its software.
  5. There is already a website version of the voting software http://snf-59337.vm.okeanos.grnet.gr:3000/ :-)
fotos commented 11 years ago

@vvatikiotis no need to update anything, the vote is still open. We are waiting on @chief unless we don't want to wait and then the votes are 4-3 for low-tech.

We had a small conversation with @vvatikiotis and @pagojo this morning regarding this issue. I was assured that @pagojo will lead it no matter the decision reached here. But please don't jump into conclusions prematurely about what will be done. It doesn't matter if the tech is available or not. This is a decision on the style of the voting we want (low-tech vs high-tech).

Some more thoughts:

  1. I didn't propose to reinvent anything, I said we'd use the existing app in my description above, read the 2nd point of Idea 2
  2. No counting necessary as @apantsiop also clarified (and we can also do @damphyr's suggestion)
  3. The person who is going to lead this (@pagojo) will have to care of everything (think out the process, possible outcomes, risk-manage it, take into account the time constraints, find the backdrop, styles, aesthetics, coordinate the mobile app and the web app, etc.)
  4. If "the WiFi doesn't work"? There is no such option in my mind. Especially if I am (or we, since this is not personal) about to find out about it Friday morning. A thousand things can go wrong, that's why good managers plan ahead and lock Murphy in the basement! And that's why we pay after all.
  5. Ditto for the server / software setup (minus the pay, unless we have to in order to guarantee this).
  6. We can prepare the lanyards and have the tickets ids && names printed (or written) on them beforehand. I'll discuss it a bit more with @nolamesa who runs the venue setup and see if it is viable (or even desirable).

I'm ending my participation in this discussion with these closing remarks: at this point, since I made the suggestion for both ideas, I don't care what's gonna be done (tho my vote stays the same) as long as it's done thoughtfully and diligently by the responsible person. The only thing I won't tolerate is a half baked solution.

chief commented 11 years ago

sorry guys now I finished with lanyards design and I will appreciate if I can vote tomorrow morning

vvatikiotis commented 11 years ago

I just reread rizopoulos idea. And it kicks imo, you cheeky bugger. If we go for the lo-tech option, it offers:

Cons

vvatikiotis commented 11 years ago

Darn. Formatting is lost...

vvatikiotis commented 11 years ago

I'll rewrite/format tomorrow. Switching off now....

vvatikiotis commented 11 years ago

I just reread rizopoulos idea. And it kicks imo, you cheeky bugger. Take a minute to read it, it's 5 lines.

If we go for the lo-tech option, it offers:

Cons