Google Blogoscoped

Forum

Click Survey  (View post)

Corsin Camichel [PersonRank 10]

Friday, July 14, 2006
11 years ago7,891 views

Pretty nice. Working on your PhD? :)

Splasho [PersonRank 10]

11 years ago #

The trouble is I think people will think too much about trying to be random. Personally I would trick people into thinking you're not interested in where they click ('click the image below to continue to the page').

iZeitgeist [PersonRank 10]

11 years ago #

Great stuff, you need many clicks though...

...Dugg: digg.com/design/Hot_Zones_Rese ...

;)

Splasho [PersonRank 10]

11 years ago #

It's a bit early for a digg, I'd wait 'til all Blogoscoped readers have done it.

iZeitgeist [PersonRank 10]

11 years ago #

What this need is as many cliclks as it can get, making it accesible to the masses just give it more randomness. I would rather have it clicked more by diggers thant by GB readers, just because the sampling is better with Digg.

>> The trouble is I think people will think too much about trying to be random. Personally I would trick people into thinking you're not interested in where they click ('click the image below to continue to the page').

I agree, but this will fade and tend to get more random as more people cick.

It would have been better just to show a white blank page and ask for clicks, that way people are not tempted to click here or there, and all the pixels get equally the same weight to get clicked!

akif sever [PersonRank 1]

11 years ago #

:D nice idea

Eytan [PersonRank 0]

11 years ago #

God bless the internet. Is it just me or does there seem to be way to many nipple-clicks?

Art-One [PersonRank 10]

11 years ago #

I deliberatly avoided the nipple zone ;-)

I clicked the pointing finger in stead. What about you?

Drew [PersonRank 0]

11 years ago #

Really, to be honest, this is a dumb idea. You are telling people to click an image of an arrow to chart where in the arrow people clicked? For what purpose? Who cares what the hot zones of an arrow are? Pretty lame.

Splasho [PersonRank 10]

11 years ago #

"It would have been better just to show a white blank page and ask for clicks, that way people are not tempted to click here or there, and all the pixels get equally the same weight to get clicked!"

There's already one of those (I can't find the link). I don't think that's what Philipp's interested in, he wants to see where on the picture people click.

My only point re: Digg was that the people on Digg are less likely to digg it if less people have done it already so I thought it'd stand a better chance of getting to the front page if waited a bit. Doesn't matter though!

Philipp Lenssen [PersonRank 10]

11 years ago #

Drew, the arrow was just supposed to illustrate the post. You are supposed to click there: :)
blogoscoped.com/click/

/pd [PersonRank 10]

11 years ago #

Philipp, are you flitering by IP?? I could click twice.. but that was like cheating the system :)-

i clicked the cards at center of the table.. in the womans hand

Is there anyway, that once a click is completed that the image below is kinda "shaded down" like kinda grayed out and then the click points are display as more "high lighted"

I think the "Chest" of a person seems to be where the most attraction lies!! :)-

iZeitgeist [PersonRank 10]

11 years ago #

>> Philipp, are you flitering by IP?? I could click twice.. but that was like cheating the system :)-

I think he is filtering by IP as you said, I am getting "Please only click once." when I trying to click twice.

Philipp Lenssen [PersonRank 10]

11 years ago #

Yeah, you can click twice, I filter by IP.
I added a skip button so you can jump straight to the result.

I'll see what I can do to visualize multiple clicks on the exact same location...

Philipp Lenssen [PersonRank 10]

11 years ago #

Edit: I changed "Please click anywhere on the image" to "Please click anywhere on the image on the survey page" to make it more clear you're not supposed to click on the post illustration.

Art-One [PersonRank 10]

11 years ago #

It's already giving a good idea.

About filtering on IP address, that isn't quite the way to do, I'm behind a firewall and just promoted this to all my collegaues. But if filtered by ip than we all can make just one click ;-).

iZeitgeist [PersonRank 10]

11 years ago #

Art-One,

I think you're right, there is no need to filter IPs as long as we want to get as many clicks as possible, doesn't matter if one person or many did them.

Phillip this give me an idea of another click survey thingy. It should be as such:

Display a blank page and ask people 10 clicks then display the cumulated clicks by everone else. This will give far more representative results I think!

Jake's View [PersonRank 10]

11 years ago #

I clicked the red diamond playing card in the guy's right hand.

pete [PersonRank 0]

11 years ago #

i clicked on the one with boobs appearing! yeah!

john sadowski [PersonRank 1]

11 years ago #

Cool idea mr. Lessen.

Splasho [PersonRank 10]

11 years ago #

I'll repeat that a blank page has already been done, it has thousands of clicks, but I can't remember URL.

Splasho [PersonRank 10]

11 years ago #

Ah, here you are ignoro.co.uk

Sunil Nair [PersonRank 9]

11 years ago #

Interesting survey, that. Isn't Google doing some research along the same lines?

Anyway, from these prelim results, it seems that the central character gets all the clicks. Or, is it the dot on her forehead that draws attention?

Reto Meier [PersonRank 10]

11 years ago #

Philipp: A suggestion for an alternative method of representing dots in (almost) the exact same spot (say +-5 pixel radius).

Maybe start with a black image, then brighten up a five pixel radius around each click – with the level of brightness depending on the clicks in that spot. If you normalise it that 'spot' on her forhead would be look 'normal' and the whole area around her chest would all be visible – but much dimmer (less visible). Most of the background would remain black.

This would show you both the areas that attracted attention plus the specific 'points' that people clicked the most. What do you think?

iZeitgeist [PersonRank 10]

11 years ago #

Many Thanks Splasho!

Very insightful indeed, It seems it follows a gaussian distribution from every corner with a higher density from the left top corner. I wonder why people tend to look more at this particular area ..? (beside the center of course)

Maybe because Google used us to look for the first results of the query there!

=)

Niraj Sanghvi [PersonRank 10]

11 years ago #

Splasho, when I tried that page, the click that it called "mine" (i.e. the one in red) wasn't where I clicked. I clicked a bit below the text, and the red dot appeared in the upper left. Makes me wonder how accurate that thing is overall.

Still, a cool idea and a neat pattern developing.

noname [PersonRank 4]

11 years ago #

it is not so random and the results can be hardly somehow interpreted as the people are trying to click to some object (money, face ...).

You should use different backgrounds, i think the best would be the serie of few images:
1. total blank image square
2. total blank image wide rectangle
3. white image with a grid
4. image consisting of colloured circles or squares
5. this image

you should also make the image smaller, use blank background and centralise it. For sociodemografy if you have a lot of clickers, you should add age/sex/left-or-right handed question.

Philipp Lenssen [PersonRank 10]

11 years ago #

That free-style click page has a fascinating result. It didn't accept my click just now though, and said I already clicked.

I know IP-checking isn't the best way to filter users. Maybe over time I can relax the rules and allow 1 via cookie + 10 via IP (if the cookie is deleted).

Reto, I'll try a bit in that direction.

Karin F [PersonRank 0]

11 years ago #

Like Niraj Sanghvi I found that ingoro.co.uk didn't mark the spot where I had clicked (interesting I, too, clicked just under the text line and had the red spot appear in the top left area).

However, I think it would be better if you didn't let people know that you were analysing where they clicked until after they had done so. I deliberately clicked somewhere less obvious (middle of back of person on left) rather than somewhere like one of the playing cards.

Philipp Lenssen [PersonRank 10]

11 years ago #

I've changed the result image into a blue and red heat map. This better shows those spots where people clicked multiple times.

/pd [PersonRank 10]

11 years ago #

yeah this gives one the visual of the "hot zones"..

IMO, this is better!!!

Philipp Lenssen [PersonRank 10]

11 years ago #

Karin, maybe later on different hot zones emerge; those of people trying to click somewhere "less obvious", those of people clicking randomly, those of people clicking on something interesting in the painting... it might be just as interesting to see places where groups of people tried to click somewhere "less obvious". I mean you gotta click somewhere and somehow come to a decision, and that's mostly based on looking at the image (plus some noise, but we can filter the noise). Maybe if we're at 10,000 or so clicks we're able to tell more!

Michael [PersonRank 0]

11 years ago #

I don't understand the point of this exercise. Telling people to click somewhere so you can see the "hot zones" is seriously polluting the sample.

mrwendal [PersonRank 1]

11 years ago #

To be fair though – had he not mentioned the "hot zones" – - how many of us would have worked out what he was trying to do anyway?

I'm guessing the vast majority.

My suggestion is a cross between the blank page and the picture idea --- maybe have a chess-board black/white image (100 x 100??) and people will still be guided to click somewhere, but won't be as vague and uninteresting as a blank sheet?

/pd [PersonRank 10]

11 years ago #

==>>"Telling people to click somewhere so you can see the "hot zones" is seriously polluting "

I dont see that happening for a new comer.. as they only see the image. So I dont think that pollutation is in occurance.

MEa culpa, I should not have mentioned "hot zones" at all – thats like a UX'er term which slipped thu my keyboard :)-

Philipp Lenssen [PersonRank 10]

11 years ago #

To me, "hot zone" is whatever emerges from the semi-random click patterns... it's nothing predefined and nothing describing a specific image location. But yes, in every analysis of the emerging click patterns you need to take every detail of how the test was presented into account; for example, "hot" may subconsciously prime men to click on the "hottest part of the hottest woman". But there may be much stronger patterns than that independent of how the introduction was written...

NateDawg [PersonRank 10]

11 years ago #

A neat variation of this would be to track the users path where the mouse moves while their deciding where to click and produce hot zones from this.

Jay Peterson [PersonRank 0]

11 years ago #

I think I am too nomal. I was predicting Three hot spots and that exactly matched. geee

It seems Google blogscoped has more male users than female users :)

Anyways it is really fun.

Neil Shaw [PersonRank 0]

11 years ago #

How does this update? The pages include two JPEGs....so how does it record a click?

Philipp Lenssen [PersonRank 10]

11 years ago #

Whenever you click on a position within the first image (and you didn't click before that day), your click position is recorded in the database... the second image you'll be seeing is a rendering of that data, using soft (almost transparent) red dots on a blue background. So the brighter the red, the more people clicked on this zone.

Wow. 39,000 clicks so far...

Philipp Lenssen [PersonRank 10]

11 years ago #

I've included a legend and another highlight color, yellow.

Ionut Alex. Chitu [PersonRank 10]

11 years ago #

You could put more images and include a link in the navigation bar. More people, more randomness.

Jake's View [PersonRank 10]

11 years ago #

i like the new version better. Good work, Philipp!

.skye.high. [PersonRank 0]

11 years ago #

pretty cool!

timmay [PersonRank 0]

11 years ago #

I think this is a brilliant idea – more should be done.

Ionut Alex. Chitu [PersonRank 10]

11 years ago #

Timmy

Pigpen [PersonRank 0]

11 years ago #

Observed that NateDawg had been here and I agree that mouse tracking would be a cool idea.

This thread is locked as it's old... but you can create a new thread in the forum. 

Forum home

Advertisement

 
Blog  |  Forum     more >> Archive | Feed | Google's blogs | About
Advertisement

 

This site unofficially covers Google™ and more with some rights reserved. Join our forum!