Greedy People

CaTaClYsM · Post by **CaTaClYsM** » Thu Mar 18, 2004 3:36 am

Seriously. WMM is going to "hold anyone back from greatness". If a video is good, people will let you know. Doesn't matter HOW many people like it, just that the people who HAVE seen it enjoy it.

TaranT · Post by **TaranT** » Thu Mar 18, 2004 3:51 am

dwchang wrote:
trythil wrote:No, that's a simple statistic fallacy. What happens if I sample all my data from one source? Say, for example, I get 30 opinions with all 10's...
But how could one sample another set of data? We're in a community and thus by definition our data is already biased to a degree yes? It's not a fallacy that with more data points, your data becomes that much more "clean" and "accurate." I mean sure two points can form a line, but obviously if 9 points are all in conjunction with the slope, that means the line is that much more accurate (ugh I'm inadvertently talking about work here >_>)

I think trythil's point - not to mention my silly proverbs

- is that there is no quality control of the input data. 100 bad measurements is no better than one bad measurement. The average is still "wrong".

For example, ask 100 people what is the correct way to score Effort. You'll probably get 100 methods, and even those that mostly agree will still score differently in specific cases. Then spread this effect over the other ratings. Without any consistency in scoring, comparing average scores is a waste of time. Whether you've got 10 scores or a 1000 doesn't make a difference. Garbage in -> garbage out.

The opinion pages are a nice way to provide and get feedback on a one-to-one basis. And the numbers help in communicating between viewer and creator. It's the aggregate (i.e. averaging the thoughts of X people who are thinking differently) that has no objective meaning.

That doesn't mean they're useless. Whether something is useful depends on what you want, and if you want some light entertainment, then comparing scores and star ratings can be fun. I know it works for me.

If you want to assume that most (however many that is) viewers are thinking in roughly (?) the same way - and this assumption seems to be unprovable - then comparing scores and ratings might make sense as long as some wide allowances are made. For example, only looking at scores that come from large samples. Or assuming that all averages between 4 and 5 are basically the same.

FallenPhoenix · Post by **FallenPhoenix** » Thu Mar 18, 2004 4:01 am

Guh. You people use too many words. 8)

To me, the higher the number, the more people liked it. There's really no room for more thought on the subject. If I really wanted to know precisely what people like and dislike about it, I'd read their opinions, however many or few there may be. And I figure that if there were something really worth noting about my video (something really good, or something really bad), then people would put it into an opinion.

dwchang · Post by **dwchang** » Thu Mar 18, 2004 2:49 pm

TaranT wrote:I think trythil's point - not to mention my silly proverbs - is that there is no quality control of the input data. 100 bad measurements is no better than one bad measurement. The average is still "wrong".

For example, ask 100 people what is the correct way to score Effort. You'll probably get 100 methods, and even those that mostly agree will still score differently in specific cases. Then spread this effect over the other ratings. Without any consistency in scoring, comparing average scores is a waste of time. Whether you've got 10 scores or a 1000 doesn't make a difference. Garbage in -> garbage out.

The opinion pages are a nice way to provide and get feedback on a one-to-one basis. And the numbers help in communicating between viewer and creator. It's the aggregate (i.e. averaging the thoughts of X people who are thinking differently) that has no objective meaning.

That doesn't mean they're useless. Whether something is useful depends on what you want, and if you want some light entertainment, then comparing scores and star ratings can be fun. I know it works for me.

If you want to assume that most (however many that is) viewers are thinking in roughly (?) the same way - and this assumption seems to be unprovable - then comparing scores and ratings might make sense as long as some wide allowances are made. For example, only looking at scores that come from large samples. Or assuming that all averages between 4 and 5 are basically the same.

No I know what he was saying, but regardless of if you have 100 "bad" data points, let's say you have 200 opinions? Are we supposed to say 50% of your data is "outliers" or "bad?" That doesn't seem right.

I mean I see your point about how different people grade differently, but that's why things are averaged. For you to say one data point is "invalid" is more or less saying that that person scores incorrectly and that their opinion is worthless.

Now don't take this the wrong way (well not you since you brougt it up, but others), but I can see what you mean in that some folks give all 10's all the time and some rate with a 5 as average (vs. 7). However, regardless of these points, it's still valid data and it's still their opinion.

I guess I'm just arguing with the fact of calling it "invalid" when it is their opinion and the fact we all grade differently. The average is taken and more or less it weighs itself out. I imagine that takes care of some (but not all) of it. There's no such thing as a PERFECT scoring system or an accurate score with respect to these things.

At the same time, the argument could be made that no video is accurate in results until *EVERYBODY* who saw it, scores it. That would be the only way to get "accurate" results I guess...*shrug*

CaTaClYsM · Post by **CaTaClYsM** » Thu Mar 18, 2004 4:56 pm

I'm thinking phade needs to set more concerete guidelines for posting opinions.

Arigatomina · Post by **Arigatomina** » Thu Mar 18, 2004 10:44 pm

CaTaClYsM wrote:I'm thinking phade needs to set more concerete guidelines for posting opinions.

Like the star scale?

What would happen if those who repeatedly gave straight 10s and 1s or even 8s to all the videos they downloaded were eliminated? We do that with the star scores - we stop counting those for the lists. Now imagine if those people who didn't have a wide enough variation in their average reviews scores were no longer counted. It might not hurt the single-time reviewers who give about 6 vids straight 10s or 1s, but those who do it a lot will be mute. I imagine the top 10% list would be effected as much (or as little) as the top star scale is when we stop counting those people who give the same star scores to all the vids they watch.

Would it be a bad idea to do this? What if a person only watches straight 10 vids? Well, we don't care when it comes to the star scale, so why should we when it comes to the reviews? Or what if we discount those who leave less than 20 words (or average under 50 on the 'usefulness' scale? It gets messier that way.

Just making the review rules the same as the star rules would cut out those who repeatedly give the same scores. I'd be interested in seeing the results of that. But it might mean a lot of coding since you'd have to make a distribution like you have for the stars 'given' list.

CaTaClYsM · Post by **CaTaClYsM** » Thu Mar 18, 2004 11:40 pm

I don't care how much you like the video. If it's good but the video quality sucks, you don't give it a 10 when it deserves a 6. I don't care how much you like it the first time you watch it. You wait for 4 or 5 more viewings before you declare you keep it in a constant loop when you're watching vids. It's an opinion, but there ARE things in a video that are matters of fact. Koop used HK DVD's for Euphoria, there were spots with incredible ghosting. I don't care if you jack off to the video on a nightly basis, YOU DON'T GIVE IT A 10 FOR VIDEO QUALITY. Not under ANY circumstances. And if someone DOES honestly think it's the best video quality they've ever seen, then they should have their opinions DELETED until they see some more vids, because it's not that hard to find one.

You see, certain things like video quality arent as much a matter of "opinion" as people think. I'm thinking we need an updated and more elaborate opinion guide which lays out the EXACTLY what parts are left up to the viewer and what parts arent.

Now I know that as I TYPE this people are already calling me jaded and superficial and how retarded it is to buckle down on the aspects of opinion giving so it's not so much a matter of opinion anymore. Now that is NOT what I'm saying, so don't get ready to start diging trenches for a long drawn out flame war over it. You can take down the straw man and I wont stop you from claiming your pathetic victory. What I'm saying, is that certain things like video quality are clear as DAY how they need to be rated. If it's MPEG-1 video meant for a VCD, you DON'T give it a 10. You give it a 6 if you're generous. If it's made from fansubs with logo's in the corner, you DON'T give it a 10, you give it a 1 and report the clear violation of the code of ethics to Phade so the miscreant can be dealt with.

Now when you're dealing with someone who used XVID and OGG audio in an OGM instead of an MKV, and your deciding whether or not he deserves a 7 or an 8 for video and audio quality, that's not being fair, that's being a jackass splitting hairs over a decicion that's utterly trivial, and a matter of prefrance for both you, and the creator. YOu don't start chizzling off points because he didn't make it the way you think it should be made, you rate it based on how it looked. If the aspect ratio was off horribly, then yes. If the guy didn't deinterlace it correctly, yes. These are things that cost people points. Not the the type of codec he used, or the container he used it in.

I'm not starting some holy cusade on opinion giving. And I'm not going into how you rate things like originality, or lip synching. Just the video quality.

But then again, this is just an opinion, take it for what it's worth.

Beowulf · Post by **Beowulf** » Thu Mar 18, 2004 11:47 pm

Good post. I've been saying this for a while now.

/chimes in

CaTaClYsM · Post by **CaTaClYsM** » Thu Mar 18, 2004 11:52 pm

Beowulf wrote:Good post. I've been saying this for a while now.

/chimes in

Yeah, you do have that nasty habbit of being right all the time. Welcome to the club. We'll have jackets made. 8)

mexicanjunior · Post by **mexicanjunior** » Thu Mar 18, 2004 11:53 pm

Castor Troy wrote:I'm sure MJ got popular through having a good attitude and being a cool guy in general despite the programs he had to work with.

Tis true, my videos have always sucked but my charming personality has won the masses over.

*AIMKISSYFACE*