Content-type: text/html Set-Cookie: cookiehash=D8TIX1F9GET8DML97LCWDC1UDL31CF7Q; expires=Thu, 17 Dec 2020 00:00:00 GMT; path=/; DMI News

DMI News

Previous Entry.. Next Entry..

Things that didn't work: Part 2

February 02, 2009 07:30

Another project that has been ongoing for years and has never been implemented is the Marquee Sign. It's not a very complicated project. In fact, several websites have done it before. I remember back in 1997 Netscape had one in their offices and I would send them daily messages to please hurry up and get around to adding CSS functionality to their browser. IE had it, but Netscape didn't and I needed it because of the Framemaker conversion from document to webpage used CSS heavily. That's neither here nor there though. Point is, they had a marquee sign you could send a message to way back when. Others have had them too. Only I was the odd man out on this one.

I could have just purchased a sign. Worst case I could have bought one for about $150 and hooked it up and had it working in 10 minutes. That's not really my style though, especially since it seems like a relatively easy project to build. It's just a array of lights and you rotate through them, updating each one with it's value at a speed fast enough to not be able to tell that each LED is only lit for a fraction of a second. However, despite starting several times, I never managed to finish it, mostly because I ran into issues, such as flicker where I shouldn't have any. The last round which I never got around to was going to implement some darlington transistors. They might work better for this. Who knows. Someday I'll get around to finishing it.

However, there was another problem casting a shadow on the whole issue. As you all know, you can send messages from the webpage. Most of you realize that these messages show up in the chatroom, as well as other places. Slightly less well known is that those messages are also fed through a voice synth and read out loud for me to hear, as well as anyone who might be listening live to my audio. Of course, if I set up a marquee sign, it would only make sense for the messages to show up there as well. The big problem with this type of interaction is that it makes otherwise reasonably civilized and intellegent people act stupid. Once they discover that they can see their messages, they try to push the envelope for some reason. This happens on a small scale in the chatroom as people aware of this feature attempt to flood it. These tend to be the malicious folk and I have no problem banning them from the site, assuming one of the automated filters doesn't catch them first. When they discover the voice synth, they try to test it out as well. Gibberish and pasting paragraphs... and sometimes entire wikipedia articles, just to see what it sounds like. On one hand, I don't really mind if the voice synth wants to read some random webpage to me (provided it's of a civil nature), but I don't want the effort to also be duplicated in every other medium. Right now, I have filters in place that detect excessive gibberish (which is moderately effective), excessively repeating characters, words, or phrases (which is very effective), and of course profanity. It will also boot you if you try to send the same message several times in a row. As can be expected, sometimes we get some false positives. It can't be perfect, and sometimes I have to unban someone that got a little carried away with his !'s when he's explaining how much he loves the site. However, it does tend to catch the real abuse, so it's hard to ignore the utility of such a system, even if it's horribly black and white in its reasoning. It IS just a computer afterall.

What brought this all back to the forefront is the awesome LCD display that danopia set up to display webpage messages. Within about 10 minutes of him getting it working, I had to unban 3 people who ran afoul of the abuse filter when they were trying to play with the display. One user got banned when he tried to make a design on the LCD. Obviously, those who simply want to play with the toys aren't banworthy, and yet, I don't want 400 character messages full of #'s and spaces being pasted everywhere that the regular messages are. Therefore, the abuse filter needs an upgrade to take these possibilities into account. I shall add it to my ever growing list. In the meantime, be gentle with it.