[Mendel]

I'll give you an AMD example because my meager mspaint skills took too long to do even one graph and its still not too good but it should get the message across

edit: this is not to say Nvidia doesn't do it, quite the contrary!


Here each line could be interpreted as being one frame per second in World in conflict but the graph begins from an arbitrarily high number of 18!
I had no idea how skewed the graph was before I started to rescale it... Notice how they dublicated every value (two "1.2x"s etc )
Think of it this way: the difference is actually 1.2x but the graph makes it look like 3.0x!



now here if you look at it closely this graph begins from 0 and each line represents one frame per second in world in conflict and the scale is much closer to what it should be in a full scale graph that is meant to give you an idea of what the differences in performance actually are, percentage wise.


Now skewing the graphs like they were in that original graph might make it look like the marginaly better card was a whole heck of a lot better even though the data indicates it isn't really...

That doesn't serve the customer, it doens't serve the ihvs and if you ask me, we gotta make it stop!

[Mendel]

Just to balance things, here's an nvidia example that I kinda criticized earlier in other thread.



Right, notice anything wrong here?

Why not go all the way, why not say just *beep* you customers and do this


edit:
(imageshack seems to have removed this image there should be a version of the above image, skewed a lot further still to make a point)

edit2: for some reason it seems the image is back!

[Mendel]

Now what do you think?

[ChrisRay]

Honestly if the reader actually pays attention to the data represented. It's not that big of a deal. I dont like percentage graphs that dont represent the actual data at the same time though.

Chris

[CruNcher]

Yeah you really have to belive their Marketing guys are super crazy or all psychologists, the truth is some really let get themselves disturbed by this and accept it as the truth without reading it carefully,also alot of times @ such Presentations you don't have time to read it carefully as the next slide is coming very fast, and so you only see the graphs.

But really funny is if their Partners then do the same thing for example Leadtek (and especialy the Kids fall for that) hehe the longer the better *cough*

I would sue them if i would be American should be easy to get them in trouble in the US Law system with this, companies get sued for much stranger things than this in the US hehe.

[Geo]

Originally Posted by ChrisRay

Honestly if the reader actually pays attention to the data represented. It's not that big of a deal. I dont like percentage graphs that dont represent the actual data at the same time though.

Chris

C'mon, now, that's a bit of a cop-out. There is a reason PR types use this kind of graph, and it's because they are specifically trying to get a "Wow!" reaction out of the unwary. That not everyone falls for the conman's rigged game does not in any way excuse the conman.

[John Reynolds]

Originally Posted by Geo

C'mon, now, that's a bit of a cop-out. There is a reason PR types use this kind of graph, and it's because they are specifically trying to get a "Wow!" reaction out of the unwary. That not everyone falls for the conman's rigged game does not in any way excuse the conman.

Exactly. And not just companies engage in these games, I've also seen web sites do likewise.

[nutball]

Originally Posted by John Reynolds

Exactly. And not just companies engage in these games, I've also seen web sites do likewise.

TBH I have more of a problem with websites doing it than the IHVs. The IHVs are marketing a product, we expect them to tell border-line lies. Websites are supposed to be there to keep them honest. If they don't, who will?

[Mendel]

Just to note, it seems like the second image of the second post is no longer available in imageshack, I wonder what happened, maybe someone reported the image to imageshack? Maybe it expired? Hope I didn't break any law or rule or anything...Maybe I should only use photobucket in the future.

edit: oh, it's back

[Putas]

Originally Posted by Geo

C'mon, now, that's a bit of a cop-out. There is a reason PR types use this kind of graph, and it's because they are specifically trying to get a "Wow!" reaction out of the unwary. That not everyone falls for the conman's rigged game does not in any way excuse the conman.

It does. People should learn to read the axis, as they do with other basic things.

[Sxotty]

Originally Posted by John Reynolds

Exactly. And not just companies engage in these games, I've also seen web sites do likewise.

That is very bad when that happens.

Originally Posted by nutball

TBH I have more of a problem with websites doing it than the IHVs. The IHVs are marketing a product, we expect them to tell border-line lies. Websites are supposed to be there to keep them honest. If they don't, who will?

This is why I jumped on digi in the other thread for saying 7% is 7%. I want a base value so I know what it means.

Originally Posted by Putas

It does. People should learn to read the axis, as they do with other basic things.

Yes people should learn, but communicators should not try to lie or obfuscate the point either by making misleading graphs. As I linked in the other thread.

Tufte
is all kinds of crazy about stamping out such things, though I disagree with some of his assertions he definitely has many valid points.

[ChrisRay]

Originally Posted by Geo

C'mon, now, that's a bit of a cop-out. There is a reason PR types use this kind of graph, and it's because they are specifically trying to get a "Wow!" reaction out of the unwary. That not everyone falls for the conman's rigged game does not in any way excuse the conman.

I dont see how its a cop out. I dont think it's a big deal. When websites start displaying graphs in this form then I'll be concerned. I dont have much sympathy for people who dont do research before they buy things. Most of these presentations ((Not all of them)) are displayed to editors and press. And editors should have enough common sense to seperate fact from bull shit. If they cant then they shouldnt be in this business to begin with.

If when the new cards are released. And anandtech, hardocp, tomshardware, beyond3d, extremetech, techreport all start using similar graphs. Then I'll be concerned. If a person walks into a store and sees that leadtek box. And makes a purchase decision off of it. Then they probably werent all that concerned to do any real research to begin with. I dont know about you guys but I learned how to read graphs in elementary school. And I'd like to assume most people spending 200-300 dollars on a graphic card can read a graph and understand the data represented. I dont like it. But I really dont think people are "That" stupid to where they cant recognise whats being displayed.

Chris

[Geo]

I think it gets used because it *works* --the confuseable are confused. I've certainly seen those kind of graphs on boxes, not just ed day presentations. To throw the easily misled under the bus with an airy wave of the hand that they deserve no better is much too Hobbesian for my taste. YMMV.

Edit: After all, one of the great articles of faith in the graphics market, "the halo effect", is built entirely on the idea that a large part of the market isn't paying all that close attention and can be influenced by factors that really have nothing to do with the purchase at hand.

[Putas]

Originally Posted by Sxotty

Yes people should learn, but communicators should not try to lie or obfuscate the point either by making misleading graphs. As I linked in the other thread..

I repeat - there is nothing confusing on the graphs, axis is clearly described.

[AlphaWolf]

Originally Posted by Putas

I repeat - there is nothing confusing on the graphs, axis is clearly described.

If there's no attempt at obfuscation why don't they start at 0?

Because there is an attempt to draw attention to whatever advantage how ever small it may be. People quickly glancing at the actual graph in full scale might be confused into thinking that there is no discernible difference or that the difference falls within the margin of error perhaps?

[Geo]

Originally Posted by Putas

I repeat - there is nothing confusing on the graphs, axis is clearly described.

So someday when some joker does a graph in millionsth of an FPS, that'll be just dandy so long as it's marked correctly. There is a reason why "attempted murder" is a crime you know.

[Sxotty]

The axis isn't clearly labled.

It says 1x in the first chart.


They should state it is a ratio of X to Y and there should not be any Y on the chart in that case.

The goal is to mislead people or they would provide a chart with actual fps. It would not be more difficult to do so. If you can present more information and have the graphic as clear or clearer you should if your goal is to communicate to the reader instead of mislead the reader.

[nicolasb]

Originally Posted by Putas

I repeat - there is nothing confusing on the graphs, axis is clearly described.

The obvious question is: if graphs like this aren't actually able to mislead, why do marketing people use them? If it made no difference, they'd have to nothing to lose by presenting the data authentically. The simple fact that misleading graphs are used as widely as they are proves beyond any possible doubt that marketing people believe they work. Maybe you think you know more about marketing than they do, of course....

Generally speaking, if you lie to somebody and successfully deceive them, the blame lies on you for being deceitful. It is not acceptable to say "well, it's their fault for believing me"; to do so would make about as much sense as saying that, if you shoot someone, you are blameless and the shooting is entirely their fault for not getting out the way of the bullet in time.

[Putas]

Originally Posted by Sxotty

They should state it is a ratio of X to Y and there should not be any Y on the chart in that case.

what? - why Y in 1d graphs?

Originally Posted by AlphaWolf

If there's no attempt at obfuscation why don't they start at 0?

Because there is an attempt to draw attention to whatever advantage how ever small it may be. People quickly glancing at the actual graph in full scale might be confused into thinking that there is no discernible difference or that the difference falls within the margin of error perhaps?

Yes the point is to draw attention to differences which may otherwise may be neglected, although they are higher then margin error.

Originally Posted by Geo

So someday when some joker does a graph in millionsth of an FPS, that'll be just dandy so long as it's marked correctly. There is a reason why "attempted murder" is a crime you know.

There is reason why suicide is not murder.

Originally Posted by nicolasb

Generally speaking, if you lie to somebody and successfully deceive them, the blame lies on you for being deceitful. It is not acceptable to say "well, it's their fault for believing me"; to do so would make about as much sense as saying that, if you shoot someone, you are blameless and the shooting is entirely their fault for not getting out the way of the bullet in time.

Your fantasy went too far.

[AlphaWolf]

Originally Posted by Putas

Yes the point is to draw attention to differences which may otherwise may be neglected, although they are higher then margin error.

Because they (or you) say so? I would prefer to judge that for myself. Sorry, but its just an attempt at deception, nothing else. Its become typical marketing and we have come to expect it, but that doesn't make it right.

[Putas]

I repeat - there is nothing confusing, decieving, lying on the graphs- because axis is clearly described.
In general it does not even have to be marketing. You don't always want 1:1 comparsion.

[ShaidarHaran]

Originally Posted by Putas

I repeat - there is nothing confusing, decieving, lying on the graphs- because axis is clearly described.
In general it does not even have to be marketing. You don't always want 1:1 comparsion.

You must work in marketing.

[AlphaWolf]

Originally Posted by Putas

I repeat - there is nothing confusing, decieving, lying on the graphs- because axis is clearly described.

BS.

In general it does not even have to be marketing. You don't always want 1:1 comparsion.

Yes, you don't want a 1:1 comparison when you want to deceive the consumer. I think we've established that. There's a number of ways they could accurately show the data so it would clearly demonstrate the actual advantage at a glance, but choose to go with a skewed graph.

[ninelven]

Just use pie charts, they auto balance.

[Davros]

One of my pet peevs is using graphs that dont start at zero