Best Voting System (How to minimize cheating)?

LineItemVito's picture

(Sorry, edit of typo in RED, below)

If you were to design a top-list voting system from scratch, one that was designed to minimize / eliminate cheating, what would you do?

Please refrain from posting notes that say, "You can't make a perfect system." I know that, you know that, everybody knows that. Nevertheless I'm interested in seeing if we can do it better.

Keep in mind that tighter security *is* possible -- look at online banks. They have figured it out (mostly -- if you don't count identity theft). Look at PGP systems. They have figured it out (mostly).

So, what would you do?

Thanks,
Eddie

LineItemVito's picture

Best Voting System (How to minimize cheating)?

(Sorry, edit of typo in RED, below)

If you were to design a top-list voting system from scratch, one that was designed to minimize / eliminate cheating, what would you do?

Please refrain from posting notes that say, "You can't make a perfect system." I know that, you know that, everybody knows that. Nevertheless I'm interested in seeing if we can do it better.

Keep in mind that tighter security *is* possible -- look at online banks. They have figured it out (mostly -- if you don't count identity theft). Look at PGP systems. They have figured it out (mostly).

So, what would you do?

Thanks,
Eddie

n/a
spargs's picture

I guess the best way to minimise cheating is to have some sort of registration system you have to go through before you can vote where you have to validate your email address. The problem is, people are far less likely to go to all that trouble just to give you a vote.

But it would be a system less open to cheating. Somewhat.

Another option might be a "enter the letters you see" box used to stop automated registrations. It might not stop cheating, but it might slow cheaters down a bit.

n/a
rabbitpie's picture

The question I have in mind is whether there are ways to minimize cheating without having the voter do something extra. Now, usually, the voter has to click an additional button, and that's cool. But I wonder how we'll all take it if we needed to do one of those capcha things or some other thing that requires us to---gasp---think.

I forgot where I saw this from... but there was a voting-esque site that simply had two buttons on the comfirm vote page, and the correct button was random. How hard would it be to defeat this system?

n/a
GregC's picture

But then you have someone with a bunch of friends that never read webcomics that will go online and vote for him just because they know him and have access to a bunch of computers/IPs/whatever. Make them log in and they'll find a way to beat that too. Any system can be beat. The ones out there aren't so terrible, they have a nice balance between security and ease of use. It's the concept that a vote equals a satisfied reader that's false. You want to assure placement with your advertising, buy a cheap ad somewhere. You want a free ad, such as a toplist ranking, you take your chances where it lands.

n/a
Kiba's picture

Remember:

All softwares have bugs. You simply cannot make a perfect system where it can't be cheated.

Halley's picture

Re: Best Voting System (How to minimize cheating)?

Acually there IS a perfect voting system...

And this is IT!!

n/a
LineItemVito's picture

If you're not serious, then LOL!

If you are serious, then you should read Alexa's FAQ which says: "Alexa's traffic rankings are based on the usage patterns of Alexa Toolbar users".

Which means, the statistics are gathered from a subset of web users. A subset that is called a "self-selected sample" in statistics. Meaning, only the web users who are geeky enough to install and use the Alexa Toolbar are counted. Think possibly the stats may be somewhat skewed towards the sites of geeky interest? Oh and by the way, the Alexa Toolbar isn't available for Macintosh platforms. Another skewed result. Oh and by the way, search for "webcomic" or "webcomics" in Alexa's website directory and you get NO RESULTS.

Again, I say LOL!

n/a
Gianna's picture

A good idea could be to have those boxes with random letters that people need to type, when someone confirms their vote. That way you'd know that it's actually a human voting and not a script.

n/a
YIRMUMAH's picture

Yeah, Alexa has to be jacked up-- I was ranking like 60,000 something, and I made a change to my page, and installed a new theme for Wordpress blog, then BAM, for some reason we went to the 300,000 rankings.

It's a really flawed system.

GooglePagerank is pretty tight though. http://www.checkpagerank.com/ -- it's how you rank through searches and links online-- and they update it often to take out link farms, etc.

I think forum voting is the best way, because you have to be registered, and once you've put in your vote, it's pretty easy to track IPS and faked IPs if anyone is paying attention. Each registered username is only allowed ONE vote. See? It would take a TON of effort to register new members just to vote.

Bluezombiesyl's picture

Onlinecomics.net had the best system for a while - you had to register and log in to vote, and you could only vote once for a comic, not like the every-day mad rush for votes on some other sites I've seen. Genres and general comic listings can be viewed a number of ways - by # of fans, by popularity(their popularity algorithm is a bit off though - you'll see comics with 2 strips and 9 fans beat a comic with 100 strips and 300 fans in popularity occasionally).

Things changed somewhat though - originally if you took out a sponsor ad for a genre, your thumbnail would be rotated on the main site as well as in an unobtrusive bar on the genre pages, but they re-vamped the sponsor ad system recently. Now your actual listing gets moved up in the genre/general comics listings so that sponsored comics appear first (in blue) before all the non-sponsored comics - and seeing the attention span of readers/surfers most people don't make it past the first few pages, so non-sponsored comics sometimes never even get a chance, despite their popularity/# of fans. I don't mind the thumbnail ads garnering more attention, but the integrated listings make it seem like the sponsored comics are actually better in quality than the others, which is just unfair.

Once our sponsor ad disappears I am pretty sure that we won't be getting the same number of votes, so it's starting to feel like we're paying for votes, rather than getting them through our own merit. What I would really like is to have the old oc.net - they didn't discourage cheating as much as we'd like, since at one point we found someone had created 10 test accounts all named "Test1" "Test2" etc and thus stacked the votes - but it still felt like a much more solid system than the other ranking sites out there. People are always going to cheat, and we constantly discourage cheating on our site since we really want to keep things merit-based, so I think sticking to places like oc.net and just refusing to participate in places like buzz/topwebcomics works for us, for now anyway.

nifboy's picture

Personally, I think the fastest, easiest, and somewhat fair and/or cheat-proof way to *approximate* "votes for a comic" is to count the number of users on its forum, and then not say that's how you got your rankings. Anything else is prone to either very low vote totals due to active participation in voting being a requirement or intervention by a spammer or author ("Vote for me!").

Aleph's picture

People who host their own comic would have access to an infinite number of emails and could create as many forum members as they wanted.

An automated system that cleaned out accounts which were only activated once, or which had duplicate IPs as well as duplicate email domains, would help to minimize cheating. Or at least make cheaters work a lot harder to cheat.

nifboy's picture

That's why I said "not tell anyone how you get your results." Until somebody figures it out and does something about it, your results are "clean" (or as clean as any internet-based result gets).

timtylor's picture

[quote:09167cea38="LineItemVito"]If you're not serious, then LOL!

If you are serious, then you should read Alexa's FAQ which says: "Alexa's traffic rankings are based on the usage patterns of Alexa Toolbar users".

Which means, the statistics are gathered from a subset of web users. A subset that is called a "self-selected sample" in statistics. Meaning, only the web users who are geeky enough to install and use the Alexa Toolbar are counted. Think possibly the stats may be somewhat skewed towards the sites of geeky interest? Oh and by the way, the Alexa Toolbar isn't available for Macintosh platforms. Another skewed result. Oh and by the way, search for "webcomic" or "webcomics" in Alexa's website directory and you get NO RESULTS.

Again, I say LOL!

On top of that, the toolbar is often classed as spyware, so it's not likely to count many security or privacy-sensitive surfers.

timtylor's picture

[quote:01889a7f29="nifboy"]That's why I said "not tell anyone how you get your results." Until somebody figures it out and does something about it, your results are "clean" (or as clean as any internet-based result gets).

But nobody's going to trust them if you don't say how you got them.

Thu, 01/26/2006 - 18:52 — nifboy
nifboy's picture

[quote:b4622b6f5b="timtylor"]But nobody's going to trust them if you don't say how you got them.

The Webcomic List hides how it gets its rankings. It gets weird results though; Filler strips for El Goonish Shive is #63 right now.

Fri, 01/27/2006 - 08:10 — timtylor
timtylor's picture

[quote:dc5cad4c0c="nifboy"][quote:dc5cad4c0c="timtylor"]But nobody's going to trust them if you don't say how you got them.

The Webcomic List hides how it gets its rankings. It gets weird results though; Filler strips for El Goonish Shive is #63 right now.

Two good reasons not to trust 'em, then. Data's like hamburgers - don't touch it if you don't know where it's from.

Fri, 01/27/2006 - 10:46 — Aleph
Aleph's picture

Plus, unless you applied some pretty funky math to it, it would be pretty easy to see over time that your forum's growth was directly tied to your ranking... even if you didn't tell anyone it would get out pretty quickly.