I've got a musician for a client. We're trying new stuff to increase interaction with the audience during and after the show.
She would love to get more in touch with her audience after the shows and she has this idea of creating a community outside of the existing social media channels. She wants people to share experiences and feelings based on the music she makes and eventually increase the engagement of the audience with the music and the music with the audience. An example is displaying comments during shows on the song she is playing right at that moment. Comments made by people who are in the audience at that right exact moment.

The problem is however, she is dead against critical or bad mouthing comments from your average internet troll. She wants it all to be peace and love.

My first proposal for a solution was introducing an administrator. A physical person that filters through every visitors' submissions and comments, but for several reasons it is not the solution we're looking for.

A follow up solution was to let the community be their own administrator. Set down rules and let people downvote ignorant or negative comments which would result in the comment being removed. The problem my client had with this solution is the fact the comment will be posted and will be visible, however briefly.

An other extreme solution that went with the whole subject of being innovative was making an algoritme that could judge whether a comment is suitable for placement. An artificial intelligence if you will. But being realistic, this idea is just impossible at the moment.

So now I'm leaning towards having no user input in the form of comments or written text what so ever. Just thumbs ups, likes or favorites. But this will limit us in creating an active community.

An other idea playing on my mind is coaxing the user into giving positive comments. Like asking them to share happy or sad memories they have with certain songs. It would all be about asking it the right way.

Any thoughts on the subject? Or any experience on the matter of keeping comments positive?

  • 33
    "I really want real feedback from my audience...but only the good feedback" - classic artist behavior. Or as I like to call them, 'sycophant magnets'.
    – Gusdor
    Commented Jul 30, 2014 at 10:24
  • 16
    If you're going to filter out anyone who says something negative, no matter how legitimate (face it, this isn't about people who put cusswords in their comments) then you might as well just think up the comments yourself and some realistically sounding usernames. That's about the same level of integrity.
    – user52403
    Commented Jul 30, 2014 at 11:10
  • 6
    Do you even know there is a problem with trolls? You might be hunting for solutions to a problem that doesn't even exist.
    – JonW
    Commented Jul 30, 2014 at 11:14
  • 5
    I'm sorry for the negativity, but I believe that "dressing" the comments form with stuff like: calling it "how are you feeling/share your experience", separate "positive/negative" comment boxes or filtering bad words (including l33t) or any other such thing will not help. If there's a text box, someone's gonna write "pewwwp" in it at some point. Commented Jul 30, 2014 at 12:53
  • 2
    I understand it seems arrogant to filter out negativity, but if it's during a show, presumably the vast majority of the attendees are fans; why do they want to read "ur fat and untalented" comments? It'd kind've ruin their buzz.
    – coburne
    Commented Jul 30, 2014 at 18:24

9 Answers 9


Hopefully my experience will be of some help, although it does involve social media.

I used to work for a large music venue, we had a huge cinematic display screen and one of the interaction ideas we came up with was to get users to submit comments to Twitter using a particular hash-tag.

The inevitable concerns were realised. How do you stop profanity, racism and general abuse? Generally these items are fairly easily filtered out using simple algorithms, the more difficult ones to suppress were comments like:

Just watched who were absolutely rubbish

The food in this place is an absolute rip-off

The security guards here are rude and unhelpful*

Now many of these comments may be fair and true and there's nothing to stop people having them on their time-lines, we just didn't want them in 1 foot letters on our screen.

In the end we went with a custom built solution that did a keyword algorithmic search. Swearing and racism were easily filtered. Pairing keywords like "rude" & "security" would also knock the comment out.

We also knocked anything out that contained "l33t speak" or simple number replacement in words.

It takes a bit of fine tuning to achieve a balanced target but it can be done.

  • Awesome to see such a similar case with filtering comments by using a algorithmic search. Nice to know it is somewhat doable. Commented Jul 30, 2014 at 9:47
  • 1
    Here's a presentation which I saw last year from an outfit who claim to be able to do this kind of thing, no doubt at a price (although the KNIME platform they build it on is open source). The interesting bit starts at slide 14: knime.com/files/005_kuduk_dymatrix2013.pdf
    – nekomatic
    Commented Jul 30, 2014 at 12:29
  • 10
    One thing to note: a filter serves as its own test bed. Be careful you don't let someone retry variations of their spam until they get their hateful message through. Consider being aggressive (but silent) with your lockout policy. It may help to limit the number displayed so if someone is unfairly blocked, they may think "oh, I just didn't get picked." Commented Jul 30, 2014 at 13:20
  • @JohnDeters: Sounds like that would be effective for about as long as it takes for one of those people to quickly skim the comments and realize that only the positive ones get "picked".
    – Aaronaught
    Commented Jul 30, 2014 at 15:21
  • @Aaronaught, I agree, but hopefully the concert will be over before the griefers figure it out. If a tweet is highlighted for 30 seconds, there may only be time for a hundred or so. In addition to racist and swear words, Paul might want to search for lists of negative words and incorporate many of those into the filter, too. Finally, for about $100 you can give someone a keyboard and the power of approval, and tell them "don't post any message that doesn't fit the vibe of the audience." Commented Jul 30, 2014 at 18:20

I always like thumbs up/down, especially when you can plot it on a timeline. However it could offer more depth. Have you considered offering a limited selection of emotions?

How are you feeling?

Music is intended to be evocative and the artist will want the audience to feel. The only bad emotion is bordom, generally indicating a lack of effect. Even anger can represent positive feedback. This information may have genuine value to the artist, not just a row of ego boosting smiley faces. They will get a cross section of the audience response without colloquialisms perhaps finding out what parts of the performance should be tweaked. You might extend this into a change in mood lighting giving instant feedback to the artist. It may also help the audience to inspect the performance on a level that they may not have previously.

Instead of emotions, perhaps there is a limited set of responses that are in line with the theme of the show.

  • 3
    Negative comments like "I don't like that song" are not really the problem. My client has had problems in the past with people just randomly telling her to die. But I like your suggestion of having a selection of emotions to select from. I will keep that in mind. Commented Jul 30, 2014 at 11:10

One way to filter out the bad apples might be take advantage of the disparity in motivation between fans and trolls. Get users to register (using their real identity?) for some kind of "fan club" closed community before they can comment.

  • The effort of registering and giving up some anonymity would increase the effort (and risk) for trolls and so decrease likelihood of them leaving their comments.
  • Banding the "club" as positive supporters of the artist might also make it something that the trolls don't want to associate themselves with.
  • Moderating members of the group might make it easy to kick out individuals.
  • Not "featuring" new members comments until they have been active in the community would mean you could moderate trolls before they had a chance to have a platform.

You'll lose a lot of fans who can't be bothered to sign up as well but it sounds like the volume of content isn't massively important to your client.

Of course all this depends on the profile of the artist. If she is very famous or has a dedicated group of trolls this might not work or the moderation might be too much work.

(I realise that I'm treating "fans" & "trolls" as polar opposites but the reality is likely to be more complicated.)

  • Simply hadn't thought about the fact the sign up process will repel most trolls. Commented Jul 30, 2014 at 14:45

Work with a whitelist and/or tagging system.

You create a number of words that the users can use, all of them positive. Your user cannot enter custom comments but has to work with what he is given. Do not let them make sentences or sentence fragments, because of the tendency of people to get really creative with them. For example, an early system by a kid-friendly game was revised after a teenager made an explicit sentence involving innocent keywords like giraffe and bunny.

A number of platforms work with this system. Steam has a tagging system for their games that allows people to enter descriptive tags. This very site allows people to tag using only prechosen words (although high-ranked users can create their own tags).

This also means that you can accept negative feedback, but it's simply not shown on the screen because it's explicitly marked not to appear. Another advantage is that you can use graphical methods to display the general mood of the crowd, like a word cloud or bar graph.


A way of having community pre-moderation could be:

  • put submitted comments into a queue which wouldn't go up on the screen immediately but could be viewed by users of the app or site
  • in order for a comment to go up it would have to be approved (or not rejected) by a sufficient number of other users
  • to get users to take part in moderating they could be required to make a certain number of moderation votes before their own comment(s) could get approved, and/or give the user the chance of winning a prize or similar if they make enough moderation votes
  • to ensure only people actually at the gig are taking part, show a code on the screen that has to be entered into the app or site, and/or get location info from the user's device

This won't stop individual users seeing negative messages in the moderation queue, and it won't stop bad stuff going up on the screen if the whole audience turns against the artist, but it might be good enough for most purposes.


I do not understand why the administrator is a problem, just make it so that before any thing is shown on stage, it has to be approved by the administrator first, literally anyone could be the administrator, as long as the front end for the system is simple enough. They just look through the comments approve the good ones as they come in and then those good comments are queued up for display, you could do whatever you want with the bad ones but they will not be shown to the crowd.

  • Our problem with a moderator was the lack of innovation. It might sound weird but the company I work for is always pushing the innovation envelope and this client who is a friend of ours has agreed to let us test some of our innovative technology during her shows and it's possibly a breeding ground for more innovative solutions. So a moderator seemed a little un-innovative. Commented Jul 31, 2014 at 6:44

Displaying comments during shows on the song she is playing right at that moment. Comments made by people who are in the audience at that right exact moment.

Then make it so, and restrict comments to the audience! That will prevent the average internet troll from spamming your event, he'll hardly pay for a ticket only to troll.

Technical solutions I could think of are filtering for the IP addresses of the concert hall's open WLAN routers, or (better and less restrictive) printing access codes to the chat on the tickets (maybe in a QR-code embedded URI).

The downside of this approach might be that the people who are supposed to listen to the music get more interested in their smartphones than in the musician.

  • During the show trolls won't be the problem no and as for recognizing people, we'll be able to recognize them via a sensor we developed that does a mixture of bluetooth and wifi analysis to create indoor location technology. But the scare for trolls is after those concerts on the dedicated website. Commented Jul 31, 2014 at 6:38

Don't even bother to try to make an algorithm to moderate comments by yourself - there are whole companies dedicated to this, and seems to be a really hard problem.

But, for what you say, what's the cost of having a person specifically to pick from the whole bunch of comments just the ones that are OK to be shown on screen live? You can even have some kind of repository of good comments that gets filled continuously (whether it be by a moderator, by the community through voting, you name it) and during the show you randomly pick from them.

If you want to limit it just to live comments, I think your only chance is a live moderator. But of course you can fill the blanks (if you have too few comments) with non-live comments.

So in the forums (or whichever means of communication you choose) you can initially show all the comments, but as they get down-voted you remove them, and after a comment has too many up-votes you mark it as a show-able candidate.


Kind of expanding on Nate Kerkhofs comment above, you could have a list of words that users can drag and drop into sentences similar to magnetic poetry.

Or maybe have a mad libs style interface where during the show a new sentence pops up every 5-10 minutes, then the users drag and drop from a list of words into the blanks to describe how they feel about the music and the show.

Making the interaction more like a game could help from people feeling like they are having their free speech censored because their comment isn't posted because of moderation.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.