Re: CEOP show Maddie is missing on 30th April 2007

Guest Sat 20 Jun 2015, 9:20 am

AndyB wrote:
canada12 wrote:
When I go to this page:
https://web.archive.org/web/20070501000000*/http://ceop.gov.uk/mccann.html
and I click on the April 30 link:
https://web.archive.org/web/20070513020901/http://www.ceop.gov.uk/mccann.html
I am taken to a page which was "created" on May 13
When I click on the PDF's, and download them, I can see, in my full version of Adobe Acrobat, that the pages were created on May 10 and May 11, 2007. I can see this by clicking on the History of the PDF.
If I put the UK PDF URL into the WBM, I can see it was only crawled once, on July 8, 2007.
This in itself doesn't prove anything. It just proves that WBM only captured the PDF on July 8, 2007. It completely ignored the PDF on all the dates previous to that. However we know the PDF was created on May 10, 2007 because it's embedded in the PDF data.

Here is the link to the UK poster PDF. You can look it up for yourself if you have full version Acrobat.

https://web.archive.org/web/20070513020901/http://www.ceop.gov.uk/downloads/CEOPHelpMadeleine_uk.pdf

It would have been entirely possible for a "dummy" page to be uploaded on April 30 with the URL's in place, but no PDF pages created yet. In other words, a link created, the page uploaded, but the PDF's not yet created and uploaded to the server.

The link would have looked like this on the "dummy" page:
http://www.ceop.gov.uk/downloads/CEOPHelpMadeleine_uk.pdf

The effect you're seeing is because the 30/04/2007 circle no longer points at the controversial archive. Instead it points at one taken on 13/05/2007. According to Wayback, the circle will disappear when the archive is next reindexed

The pages are indexed in a data structure called a linked list. Each item (node) in a linked list has three parts - a head, the content, and the tail. The head is the date of the previous capture, the page is the content, and the tail is the next capture. Then they are all linked, by the tail of one thing becoming the head of the other - sometimes this is referred to as a pointer.

WB have just changed the head of the May capture so that it does not link to April 30 any more. When they do a full re-index, the blue April 30 dot will indeed disappear and the May capture will become the first item in the list.

This is part of the code from the May capture that demonstrates it -

CEOP show Maddie is missing on 30th April 2007 - Page 14 Linked10

CEOP show Maddie is missing on 30th April 2007 - Page 14 Linked10

Guest Sat 20 Jun 2015, 9:25 am

Resistor wrote:
The pages are indexed in a data structure called a linked list. Each item (node) in a linked list has three parts - a head, the content, and the tail. The head is the date of the previous capture, the page is the content, and the tail is the next capture. Then they are all linked, by the tail of one thing becoming the head of the other - sometimes this is referred to as a pointer.

WB have just changed the head of the May capture so that it does not link to April 30 any more. When they do a full re-index, the blue April 30 dot will indeed disappear and the May capture will become the first item in the list.

This is part of the code from the May capture that demonstrates it -

Thanks Resistor. So now we have definitive proof that the WBM people have been at work altering the links. The question still remains in my mind, is it because WBM had a genuine error, or because they've been told to alter something that was a genuine listing.

I'm not sure we'll ever know, unless this ends up in court as evidence.

froggy Sat 20 Jun 2015, 9:28 am

Resistor wrote:To let you all understand, I will not be responding to anyone who simply sits and bleats "the date is wrong" or "it is a glitch" without offering any explanation as to why the date could be wrong, or exactly what the glitch is. That gets us all nowhere. I'm also not going to respond to goading and insults.

I will answer any questions to the best of my ability and if I don't know, I'll happily admit that I don't know. And neither will I get into any pissing competitions as to who has the biggest and best qualifications and experience, because there is no way of proving any of it on a forum, for all you lot know I could be Stephen Hawking (I'm not, but I could be :-))

What I did do last night, when I could not sleep, was trawl round a developers forum where the people who actually work on the open-source Java of the WBM hang out, and I found this.

My understanding of this is that the crawler archives pages on a certain date but then those pages can contain stuff that is crawled on a different date. In other words, it is the content of the page that might be unreliable from a timing point of view, not the date the page was found and archived.

Does that then take us back to the point that the 'McCann' page must have existed on the 30th, irrespective of content ?

Guest Sat 20 Jun 2015, 9:30 am

Yes Canada, you are quite right to ask questions and I dearly want an answer too, from a technical and professional POV. I know what I believe, though. Others are, of course, free to disagree, provided they do it civilly. I believe in laying out both facts and theories and letting others make up their own minds. The "battering-over-the-head-with-a-big-stick" method just does not work on me, in fact it makes me even more inclined to believe the exact opposite.

Guest Sat 20 Jun 2015, 9:31 am

Yes Froggy, IMO it says that the contents are more likely to be inaccurate than the date.

AndyB Sat 20 Jun 2015, 9:41 am

Resistor wrote:What I did do last night, when I could not sleep, was trawl round a developers forum where the people who actually work on the open-source Java of the WBM hang out, and I found this.

My understanding of this is that the crawler archives pages on a certain date but then those pages can contain stuff that is crawled on a different date. In other words, it is the content of the page that might be unreliable from a timing point of view, not the date the page was found and archived.

That's very interesting because it goes some way to substantiating what I was speculating about earlier. If the page is reconstructed at retrieval and some of the page content, such as the news links on the home page, is added then rather than when it was archived its possible that the different date could be in the future. Looking at the source again, I think it probably is reconstructed at retrieval. Look at this comment:



To me that says that the page was archived on 30/04/2007 at 11:58:03 then reconstructed on 17/06/2015 at 10:35:10. How else would the "RETRIEVED FROM THE INTERNET ARCHIVE ON 10:35:10 Jun 17, 2015" bit get there?

Guest Sat 20 Jun 2015, 9:43 am

I'd also like to add that I am not the only person in the world who knows this stuff. We can't all be erased or hidden or shut up. This is out there now, no matter what peoples' opinions of it, it is not going to die or just go away.

Guest Sat 20 Jun 2015, 9:45 am

@Andy, yes, the developer on the forum does use the term "replay" which tends to suggest that it is held as a series of vectors, which are then reconstructed at retrieval time, and a new datestamp added.

AndyB Sat 20 Jun 2015, 9:52 am

Resistor wrote:@Andy, yes, the developer on the forum does use the term "replay" which tends to suggest that it is held as a series of vectors, which are then reconstructed at retrieval time, and a new datestamp added.

Thanks. I might be coming back round to your way of thinking Very Happy

Guest Sat 20 Jun 2015, 10:02 am

As a complete numpty in all of this,is my understanding of what resistor is saying is this that the page was created or at least saved/archived or how ever its described at 11:58:03 Apr 30, 2007 and irrespective of when it is pulled again for editing the original date and time will remain constant?

Guest Sat 20 Jun 2015, 10:05 am

I believe so Caricature, yes. According to the stuff on developer forum I found, it seems that the contents of a page can change dynamically, as it is rebuilt by the software every time it is replayed. But the original date/timestamp has been generated by the server hardware, and is a lot more difficult to alter or be "wrong".

Guest Sat 20 Jun 2015, 10:08 am

@Resistor,thanks.

chirpyinsect Sat 20 Jun 2015, 10:09 am

Resistor wrote:I believe so Caricature, yes. According to the stuff on developer forum I found, it seems that the contents of a page can change dynamically, as it is rebuilt by the software every time it is replayed. But the original date/timestamp has been generated by the server hardware, and is a lot more difficult to alter or be "wrong".

May I ask what you think has been done to change the way we are now seeing the " Apr 30" page? Because it clearly has now been altered.
In other words they have been able to remove the whole page and replace it with a repeat of what was there on 27 Apr. Now the Maddie stuff appears on 14 May but with 2 photos and not 1 as before and there is no mention of the latest news dated Oct 07 that we all saw originally.

Guest Sat 20 Jun 2015, 10:19 am

chirpyinsect wrote:
Resistor wrote:I believe so Caricature, yes. According to the stuff on developer forum I found, it seems that the contents of a page can change dynamically, as it is rebuilt by the software every time it is replayed. But the original date/timestamp has been generated by the server hardware, and is a lot more difficult to alter or be "wrong".
May I ask what you think has been done to change the way we are now seeing the " Apr 30" page? Because it clearly has now been altered.
In other words they have been able to remove the whole page and replace it with a repeat of what was there on 27 Apr. Now the Maddie stuff appears on 14 May but with 2 photos and not 1 as before and there is no mention of the latest news dated Oct 07 that we all saw originally.

The April 30 page was originally the first page in the list. So the head would be April 30. The tail of it was May 13, which became the head of the May 13 page. The tail of May 13 in May 22, which becomes the head of May 22 page, and so on.

They simply broke the link between April 30 and May 13, and made the pointer for the very first entry point to May 13 instead.

The big calendar thing probably works on a different algorithm, an automatic one, that regenerates by itself and just shows the heads of all the nodes. The next time it regenerates, the head to April 30 will disappear and so will the dot. The head of April 30 - at the moment - is still there, but I strongly suspect the content has been whooshed. As has a lot of other content in this whole sorry mess.

chirpyinsect Sat 20 Jun 2015, 10:46 am

Thank you Resistor. Keep on truckin.

Guest Sat 20 Jun 2015, 10:47 am

chirpyinsect wrote:
Resistor wrote:I believe so Caricature, yes. According to the stuff on developer forum I found, it seems that the contents of a page can change dynamically, as it is rebuilt by the software every time it is replayed. But the original date/timestamp has been generated by the server hardware, and is a lot more difficult to alter or be "wrong".
May I ask what you think has been done to change the way we are now seeing the " Apr 30" page? Because it clearly has now been altered.
In other words they have been able to remove the whole page and replace it with a repeat of what was there on 27 Apr. Now the Maddie stuff appears on 14 May but with 2 photos and not 1 as before and there is no mention of the latest news dated Oct 07 that we all saw originally.

Sorry chirpy, I overcomplicated the last answer, because I missed that bit (still half asleep). You aren't seeing the April 30 page anymore, you are actually looking at the May 14 page, even though you've clicked on April 30 in the calendar.

Guest Sat 20 Jun 2015, 11:16 am

What use is the Wayback machine as an archive if it changes what it captures at a later date? How can there be court cases relying on this data? I don't have any quals in IT nor member if MENSA I feel so dumb!!!!! NOT

Guest Sat 20 Jun 2015, 11:22 am

Hongkong Phooey wrote:What use is the Wayback machine as an archive if it changes what it captures at a later date? How can there be court cases relying on this data? I don't have any quals in IT nor member if MENSA I feel so dumb!!!!! NOT

That disturbs me as well, HKP. But according to the post I made at 8:11 on the previous page, it looks as though that is exactly what happens. Mad

And it's clearly disturbed the author of that post too, because he is asking for that facility to be limited, or for people to be able to specify what they want included/excluded by using date parameters.

Guest Sat 20 Jun 2015, 11:33 am

A poster on CMOMM made a what to me seems a reasonably reply in this.

It is impossible for a crawler to access a page if it doesn't know it's there. There are two ways for a crawler
to know a web page exists:

1) The crawler is "told" about the web page, in other words the page is submitted to the crawler.
2) There is a hyperlink to the page somewhere on the Internet that the crawler follows.

It therefore follows that if neither of those two actions have taken place a crawler cannot know a page is
there.

Does this make sense?

They then go on to say

It's possible that someone working on the page carried out either of my points 1 and 2 above. Note though,
that's not an accident, it has to be done on purpose.

It's unlikely in the scenario we're talking about that someone submitted the web page to a crawler, so to stick
with just option 2, the hyperlink option, someone could have put a link to the page somewhere on the
Internet and then a crawler could find it.

Option 2,whistle-blower?

Guest Sat 20 Jun 2015, 11:36 am

Resistor wrote:
Hongkong Phooey wrote:What use is the Wayback machine as an archive if it changes what it captures at a later date? How can there be court cases relying on this data? I don't have any quals in IT nor member if MENSA I feel so dumb!!!!! NOT
That disturbs me as well, HKP. But according to the post I made at 8:11 on the previous page, it looks as though that is exactly what happens.

And it's clearly disturbed the author of that post too, because he is asking for that facility to be limited, or for people to be able to specify what they want included/excluded by using date parameters.

Your post at 9:11 you mean.
Why have we had these pages used as evidence if it cannot be proven that what is captured at the date and time stated was what was there. Richard D Hall is also adamant that the date is correct (claims software development as one of his former jobs)

Guest Sat 20 Jun 2015, 11:37 am

Yes, a lot of crawlers work like that, the biggest and most obvious being Google, but probably not all of them. It's perfectly possible for some to work in a different way and hoover up everything on a server, whether it has been linked to or not.

Guest Sat 20 Jun 2015, 11:41 am

Hongkong Phooey wrote:
Resistor wrote:
Hongkong Phooey wrote:What use is the Wayback machine as an archive if it changes what it captures at a later date? How can there be court cases relying on this data? I don't have any quals in IT nor member if MENSA I feel so dumb!!!!! NOT
That disturbs me as well, HKP. But according to the post I made at 8:11 on the previous page, it looks as though that is exactly what happens.

And it's clearly disturbed the author of that post too, because he is asking for that facility to be limited, or for people to be able to specify what they want included/excluded by using date parameters.
Your post at 9:11 you mean.
Why have we had these pages used as evidence if it cannot be proven that what is captured at the date and time stated was what was there. Richard D Hall is also adamant that the date is correct (claims software development as one of his former jobs)

No, 8:11 HKP. Your individual profile must be displaying the time as UTC + 1. Have a look in your profile, preferences. Mine is set to Reykjavic time, which is the same as UTC.

CEOP show Maddie is missing on 30th April 2007 - Page 14 81110