News:

Printed Amstrad Addict magazine announced, check it out here!

Main Menu

CPC Wiki is no more #1 CPC TopSite?

Started by TotO, 12:20, 25 January 13

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Devilmarkus

Ok, lets be more serious...
I really don't know now:
Should I ban his account?
The "Toplist" page was never meant to be a platform like "My di** is larger than yours" but a platform to share as much CPC links as possible.

On the other hand he is pushing his page to maximum when he cheats.

So a "fix" for all would be: All apply the same cheat on their pages and that's it...

(No, emmm... Better lets fair play I guess)
When you put your ear on a hot stove, you can smell how stupid you are ...

Amstrad CPC games in your webbrowser

JavaCPC Desktop Full Release

Gryzor

So that would be "my bot is bigger than yours"?

ralferoo

Quote from: Devilmarkus on 19:12, 07 February 13
So a "fix" for all would be: All apply the same cheat on their pages and that's it...
I guess it depends on what the metric is supposed to be. If it's unique visitors, he's clearly and knowingly cheating. To be honest, I'd be tempted to count the visit as rand()*1.9-0.9 if the rnd parameter is passed. That way his count will keep increasing and he'll be none the wiser, but won't go up as much. Or count it as 0 if rnd is passed.

It sounds like the current behaviour is supposed to check unique visitors per day, but relying on the browser doing via by an expiry time isn't great for that anyway. There are all sorts of reasons why something might count multiple times doing that, and also proxy servers can artificially reduce this count (which might actually be the reason for his cheating - when I did web development I've done a similar trick to force it to get unique data for multiple users behind a proxy).

Personally, I think a count per page is fair. If I look at 30 pages on the CPC wiki and 1 page on another forum, the CPC wiki deserves 30 times as many votes IMHO...

Devilmarkus

It's designed to count unique users.
If each pageview would be counted, the CPCWiki would have thousands of visitors daily.

The new code I applied now avoids this.

Also sad: The code-owner reads this but does not write a statement...
When you put your ear on a hot stove, you can smell how stupid you are ...

Amstrad CPC games in your webbrowser

JavaCPC Desktop Full Release

Devilmarkus

When you put your ear on a hot stove, you can smell how stupid you are ...

Amstrad CPC games in your webbrowser

JavaCPC Desktop Full Release

Gryzor

@ralferoo: actually it also counts (and lists) total views as well, but ranking is done according to unique visitors. You're not wrong in what you're saying - some sites have more content than others - but then, maybe we should start counting by time-on-site? Start weighting page freshness? :D It gets complicated... unique visits does give a pretty good idea about how popular a site is.


No, it's not ideal, but then again it's just supposed to be a quick and dirty way to list sites on a specific subject (like the old Rings - anyone remembers those?). And I guess should be based on the honour system a bit, too.


And I'm completely puzzled as to why something like that would be done.

db6128

I'm more surprised by who's doing it, seeing as it's very different from the high standard of manners presented elsewhere.
Quote from: Devilmarkus on 13:04, 27 February 12
Quote from: ukmarkh on 11:38, 27 February 12[The owner of one of the few existing cartridges of Chase HQ 2] mentioned to me that unless someone could find a way to guarantee the code wouldn't be duplicated to anyone else, he wouldn't be interested.
Did he also say things like "My treasureeeeee" and is he a little grey guy?

Gryzor

Same here for the same reason. Hope we get to hear from him sooner rather than later... :'(

Snake_Plissken

Where is the interest ?

It's not a contest.

Gryzor

It's a motivation, but certainly not a contest.

Devilmarkus

When you put your ear on a hot stove, you can smell how stupid you are ...

Amstrad CPC games in your webbrowser

JavaCPC Desktop Full Release

Gryzor


Devilmarkus

#62
Well we added a simple routine in the image code:

// Avoid GET attacks
foreach($_GET as $k => $v){
     if ($k != "u" )    {
          header("Location: hxxp://yoururl.com/yourimages/cheater.png");
          die();
     }
}


You normally embed it with variable ?u=Username.
So if a bad variable is found (e.g. ?rnd= or an additional is found: &rnd=) *zzzaP Gotcha! ;)

Of course, when Proda returns the image code to default (as it should be) the toplist counter works properly and the real hits will be counted again...
When you put your ear on a hot stove, you can smell how stupid you are ...

Amstrad CPC games in your webbrowser

JavaCPC Desktop Full Release

Devilmarkus

So lte's guess, which page is now on #1...
When you put your ear on a hot stove, you can smell how stupid you are ...

Amstrad CPC games in your webbrowser

JavaCPC Desktop Full Release

Prodatron

Hi guys, I was away for a while (sorry, didn't read this thread before right now) and I am a little bit surprised what happend.

Regarding the "cheating":
@Devil: Just check your logfiles, this random number has been added years ago, it was always present in the code, since I added the Toplist-banner in the website (already in Kangaroo-times ;) ).

Maybe you know that I am coming from the online advertisement business, and so I know quite a little bit about this website analytic stuff. Adding a random number to a counting pixel is NOT cheating, you always do it for getting more precise numbers, as you prevent proxy and cache mechanisms from loosing counts.

Please note, that there are two values in website statistics:
1.) page impressions
2.) unique users/visitors

- Page impressions count each hit of a URL of a website. In this case it is the BEST way to add a random number to the pixel. Proxies and/or caches do not always follow HTTP expire times, and that's the reason why you should add random numbers to measure EVERY page impression.
- Unique users are counted either on cookie or on IP/session base. On each impression the system checks, if it's a known user. The best way is to use cookies with an expire time, but if they are blocked, you have to use the "fingerprint" methode, which includes the IP and the browser configuration.

I am really wondering why you don't know the difference between page impressions ("Total PVs") and unique users? Unique users ("Unique PVs" how they are called in your Topsites) are NOT affected by the random number which any statistic software adds to its counting pixels (if possible) because of the reason I mentioned above. It was quite funny to read, that you thought that pressing F5 would increase the Unique PVs. I am sure that you know quite well, that this would never increase the Unique PV counter!?

I have no idea where this explosion of the page views came from. I had this serveral times in the past (but yes, for shorter times) but this is NOT a robot (and not the random number methode you mentioned as it's the usual way for statistics and has been existing all the years) Why don't you check your logfiles?? The page views went down again, so I guess it was just another crazy post somewhere or whatever.
I can remove the random number (which will not affect the Unique User counter) to make you happy again. But I would more recommend to add such thing to each code to get more precise "Total PVs" numbers for each page (which ISN'T affecting the ranking, which is based on "Unique PVs"!).

CU,
Prodatron


GRAPHICAL Z80 MULTITASKING OPERATING SYSTEM

Gryzor

Hello Prod,

Thanks for trying to clear the bad air!

A few weeks back some spammers tried to get on board here, and that inflated our stats. I blocked those IPs and the numbers went down again, but in your case it doesn't make much sense - not much a spammer can do on your site! I can see on obvious reason by googling it though, have you seen your site's analytics to see where the visitors come from?

Again, thanks for explaining a bit...

T

Devilmarkus

Yes, thanks for explaining.
I actually removed the check and added a random number to my website's button, too.
Perhaps you are right.
When you put your ear on a hot stove, you can smell how stupid you are ...

Amstrad CPC games in your webbrowser

JavaCPC Desktop Full Release

Gryzor

I would like to point out that the original question remains - where does that traffic come from? It's going into its fifth week now, so I find it very unlikely to be 'valid' traffic. Unless we're missing something in which case - as I said earlier - it'd be a nice surprise.

Prodatron

Will try to find this out as soon as possible.
I remembered, that I integrated Googly Analytics on the homepage. So I logged in right now, but the strange thing is, that it shows only 1 impression for the whole last 4 weeks, so maybe the code is outdated.
Will try to get some logfiles from my hoster.

CU,
Prodatron


GRAPHICAL Z80 MULTITASKING OPERATING SYSTEM

Gryzor

For me GA always shows numbers a bit higher than the TopSites stats :)


But since the visits seem to be still up high, just updating your GA code should be enough, even without the host stats!

ralferoo

Quote from: Devilmarkus on 12:53, 11 February 13
Yes, thanks for explaining.
I actually removed the check and added a random number to my website's button, too.
Perhaps you are right.
If you do that, you should also remove the expire time from the header or it will skew all the other results the other way.

I kind of agree with Prodatron as I've had to do similar things myself in the past to avoid over enthusiastic proxy servers (although it's always been within the domain of the web app concerned and not for statistics reasons). Really, you want to send a cookie with the image so that you can identify repeat visitors, but there's all sorts of stupid EU laws about cookies now that it's easiest not to, and to be honest, I'm not sure a client receiving a cookie from an online image would actually cache it, because normally you'd only be concerned with cookies from the page requests. Coupled with the image coming from a different domain, I'd full expect that to not work to identify unique visitors well.

But, personally, I think page impressions is a more useful metric anyway. As an aside, I've noticed that the news items pulled from Octoate.de also include a top-sites banner presumably from the RSS feed, so that's counting an impression without me even visiting the site - just from reading a page on cpcwiki! So, just clicking on one news item on cpcwiki and staying within cpcwiki gives his pages equal value to cpcwiki, even though I've spent ages on cpcwiki and not even looked at his site. I'm not saying it's wrong that he's done that, just that unique visitors is a less useful metric than page impressions.

Gryzor

Yeah - actually I discussed this very recently; the news bot is a mere rss fetcher, it doesn't read the contents of the feed. So if Octo includes the button in the feed then yes, his site gets an extra hit every time it is served from the wiki. But there's some kind of divine symmetry here :) As a reward, the better your articles the more will read you, either by visiting the site or by some other means!

Prodatron

Quote from: Gryzor on 13:43, 11 February 13
For me GA always shows numbers a bit higher than the TopSites stats :)
But since the visits seem to be still up high, just updating your GA code should be enough, even without the host stats!
Yes, it depends on where you place the code, and it is also possible, that the GA-pixel response more faster than the TopSite-image.
Just saw, that my GA-code is completely outdated, I will add a new one now.

Quote from: ralferoo on 15:00, 11 February 13
If you do that, you should also remove the expire time from the header or it will skew all the other results the other way.
I don't think so. The important thing is, that the pixel/image is loaded ONCE everytime directly from the counting-server, when the user enters one page. This can be forced with a random-value. In this case you can measure the page impressions in the right way. The expire time is just an additional option to prevent caches/proxies from passing over the server.

Quote from: ralferoo on 15:00, 11 February 13
Coupled with the image coming from a different domain, I'd full expect that to not work to identify unique visitors well.
No that works as good as for pages. It's the way how Adservers are working. They are using the cookies of their own images/count pixels, and these cookies always come from the adserver domain. This also makes it possible to track the same user over different websites. The new EU rule forces the advertisement companies to provide opt-ins or opt-outs for the users, so that they have the possibility to prevent tracking themself.

Quote from: ralferoo on 15:00, 11 February 13
But, personally, I think page impressions is a more useful metric anyway. [...] I'm not saying it's wrong that he's done that, just that unique visitors is a less useful metric than page impressions.
Both are quite interesting values. You could decribe it like this:
- the number of Unique Users (visitors) show the "quality" of a website. A high number shows, that a lot of people are interested in this site.
- the number of Page Impressions in relation to the unique users show the "quantity" of a website. A high number shows, that single users like to surf a lot over the same site in the same session.


GRAPHICAL Z80 MULTITASKING OPERATING SYSTEM

Devilmarkus

@Proda:
Well I need to apologize then.
Checked your site backwards to 2010.
Indeed there you already used the random number.
But I hope you can agree with me, that when a website stays for long time @ #1, I'll check deeper the code which is used, and when I stumble across a value which is not the standard code, I normally guess, that something is weird here....


When you put your ear on a hot stove, you can smell how stupid you are ...

Amstrad CPC games in your webbrowser

JavaCPC Desktop Full Release

ralferoo

Quote from: Prodatron on 15:36, 11 February 13
I don't think so. The important thing is, that the pixel/image is loaded ONCE everytime directly from the counting-server, when the user enters one page. This can be forced with a random-value. In this case you can measure the page impressions in the right way. The expire time is just an additional option to prevent caches/proxies from passing over the server.
I got the impression that DevilMarkus was using it the other way, so the expiry time was set to 24 hours so that the browser would only request the image once every 24 hours. I know you can use it to force instant expiry too which is what I meant by removing the expiry time.

Powered by SMFPacks Menu Editor Mod