Re: Steve Marsden's WBM screenshot: The CEOP Home page for April 30, 2007 also refers to Missing Madeleine.

by HKP 01.07.15 12:20

I can assure you there has been no attempt at anything, I have had it explained that due to a slow download of the index I was probably logging into the 'other' forum with my username. Somehow I've managed to add a url (having multiple windows open) which as I've now found out is very easy. Sorry there is no plot and I have no other explanation.

I have sent a PDF and explanation of contents to admin hopefully she will upload it for me.

by HKP 01.07.15 12:26

@syn. Are you saying you added that url????? I thought I was going bonkers.

You are also getting me mixed up with dates, the McCann . html was added 30/04 and the Madeleine jpg's added that day too.

by Syn 01.07.15 12:28

HKP wrote:@syn. Are you saying you added that url????? I thought I was going bonkers.

You are also getting me mixed up with dates, the McCann . html was added 30/04 and the Madeleine jpg's added that day too.

Yup HPK I did it, sorry, you are not going mad. I meant to post that I had done so to let you know :)

I will try and find the post where you refer specifically to the two urls I mentioned above too. Was either on the 28 or 29 June :)

by siobhan3443 01.07.15 12:31

@syn thanks, that explains that and you called the subset

fair play, it proves your point!!!!

by siobhan3443 01.07.15 12:32

@hkp

apologies

by Syn 01.07.15 12:35

siobhan3443 wrote:@syn thanks, that explains that and you called the subset

fair play, it proves your point!!!!

Thanks :) I should have let people know what I'd done but I got distracted after and forgot oops!

In Edit meant to add that I added another this morning, something along the lines of [You must be registered and logged in to see this link.]

by **Tony Bennett** 01.07.15 12:39

Syn wrote:Problem with subsets says Chris Butler....

.... that brings us right back to the beginning where I posted this on 17th June when this merry go round first started.

I have now underlined the relevant part

[You must be registered and logged in to see this image.]

Can you please help us on this?

Given that you agree (I think) with Christopher Butler that

(a) there was no 'mcmann.html' page, orphan or otherwise, on the CEOP website on 30 April, and that

(b) there has been a 'subset error' (which the above report 'deprecates') of some kind that has given this 'false' reading...

1. how many hundreds/thousands/tens of thousands/hundreds of thousands of Wayback records are false because of this faulty subset, and

2. when did Wayback discover this 'subset error', and

3. over what period did this subset error persist?

TIA

by HKP 01.07.15 12:48

@syn. I don't have time to give full reply right now but you've actually proved nothing. I will explain later when I have calmed down.

by roy rovers 01.07.15 12:56

Sometimes when I come to this website I am pre logged in and other times I'm not and have to re log in. Is this an establishment plot as this is a McCann website?

by HKP 01.07.15 13:03

@syn. You have demonstrated that even to add a url it takes the date manually or otherwise it was added ergo McCann and Madeleine added either 49/04 or before.

by HKP 01.07.15 13:09

@syn. Last post should have said 30/04.

Further this subset error I will give more info later that shows the error had no bearing on the original data captured 30/04/07

by NickE 01.07.15 13:16

Delete

by SixMillionQuid 01.07.15 13:57

Syn wrote:Problem with subsets says Chris Butler....

.... that brings us right back to the beginning where I posted this on 17th June when this merry go round first started.

I have now underlined the relevant part

[You must be registered and logged in to see this image.]

We've seen HTML print outs which show where the Wayback machine has inserted its own HTML tags. But the original HTML contents of the page being copied is not altered unless there are links to other pages that have been archived. And at the bottom of the HTML it states when the page was archived and when it was retrieved by a user. In the previous thread the mccann.html archive date was 30 April 2007.

by XXXXXXXX 01.07.15 14:36

So is syn saying that anyone can create a page and stick it in the WBM archives?

by **Jill Havern** 01.07.15 14:40

HKP wrote:I have sent a PDF and explanation of contents to admin hopefully she will upload it for me.

Yer tis:

HKP wrote:

I've added the file which shows that the Madeleine jpg's have a unique marker (as a subset) which is (s(beokrn453z22tm55hjfuox45)). On 30/04/07 104 URLs had this marker, 102 of the URLs were repeated on average 20 times, the Madeleine jpg's were not repeated at all (they are a one time capture). They also follow the same rules as the other jpg's in the main index (for all nearly 9000 URLs) that they have a single day capture (in this case 30/04/07) a few other single day captures and finally a date range capture. I consider that because of the type of file (jpg) the repetition did not occur.
To sum up on 30/04/07 (from & to) the only files which were captured and did not repeat were Madeleine 01 jpg. Madeleine 02 jpg. and that is 2945 URLs in total (remember the average of 20 repeats).

I will also add that there is one other jpg which was captured in a date range of 30/04/07 to 08/07/07 it again has no repeats.

[You must be registered and logged in to see this image.]

by siobhan3443 01.07.15 14:52

I will also add that there is one other jpg which was captured in a date range of 30/04/07 to 08/07/07 it again has no repeats

@hkp -don't leave us with a cliffhanger, which one?

what's this page showing us? i can't see the number of times things were repeated.just row labels and page names is it loading properly for me?

has skylark1 got any reply yet?

by HKP 01.07.15 15:00

[size=40]@ Get'emGonçalo[/size]

[size=40]Thanks but the important part has been cut off I.e the number of repeats[/size]

by HKP 01.07.15 15:03

@siobahn. Sorry to disappoint but its the ceop awards jpg

by siobhan3443 01.07.15 15:07

@hkp thanks, i thought something was amiss

tennis ball picture is jpeg 2 wbm shows this was also archived on the 3 july and the july from their calendar
https : //web . archive . org/web/20070701000000*/ http : // www . ceop . gov . uk/madeleine_02.jpg

note the odd date in the url

by siobhan3443 01.07.15 15:09

sorry dates it was also captured should read 3 july and 8 july 2007

by siobhan3443 01.07.15 15:24

https : //web . archive . org/web/20070901000000*/http ://www . ceop . gov . uk/madeleine_01 . jpg

jpeg 1 doesn't show a 30 april capture date on the calendar , first captured 6 june 2007

by **Jill Havern** 01.07.15 15:42

Sorry about that....it's cos I enlarged it so it could be read blushing1

[You must be registered and logged in to see this image.]

by HKP 01.07.15 15:44

@siobahn. As I explained earlier the Madeleine . jpgs were single captures on 30/04/07 then on a specified day (all past 30/04) they are included in a date range.

Madeleine 01 captured in subset (ref above. ) on 30/04/07
Madeleine 02 captured in same subset (re above) 30/04/07

Madeleine 01 captured out with subset in range starting 06/06/07 to 22/06/15

Madeleine 02 captured out with subset range 30/04/07 to 22/06/15

by whodunit 01.07.15 16:05

Get'emGonçalo wrote:
whodunnit wrote:Hello--thanks for the invite admins. I successfully registered here some months ago but never received an activation e-mail. If a mod would like to go through old registrations, iirc I registered as 'whodunit' with one n. If not let me know and I'll re-register.
I've activated your account.

Thank you!

by Skyrocket1 01.07.15 16:42

As promised here is the response from the Prof, sorry it's a bit late:

Sent: Tuesday, 30 June 2015, 18:26
Subject: Re: A bit of advice - many thanks (I'm in the UK and have a question about data capture/archiving web pages)

Dear ******,

As one of my interests is traceability of computational results as well as data management practices, I do find the situation you describe very interesting. Altough I have to admit that, at the time being, I can do little but speculate at a very abstract level.

The fact that it's only the existence of a certain page that is to be determined makes the situation quite a bit easier. But still I doubt it will be easy to trace the reasons for the mistake without detailed cooperation from the IA on the technical level.
If there is an error withthe date assigned to a crawl, then it should be a systematic error that would apply to many pages showing similar characteristics, i.e. being crawled for the first time, and all of them showing a systematic off-set to the real first crawl time by a certain delta; or all pages collected by a specific crawl job showing such a behavior, or something like that.
Manual interference can, of course, be the case as well, but it does not seem obvious to me, what kind of manual interference (which usually is necessary if a crawler gets stuck, a harddisk for a certain queue fails, etc.) would lead to such a date misalignment. I am sure there might be reasons, but I can't think of one righ away, and even that should be traceable across several pages (even unrealted ones as a crawler/job/queue is never dedicated to a specific website or web domain but usually collects pages from a myriad of domains. One could try to find pages crawled by the same job/affected by the same interference that exhibit some other means of verifying the date (such as having it explicitly stated on the webpage) to verify such a systematic date off-set error. But in any case, any such effort is likely to resemble quite a bit of forensic work, collecting material, establishing hypotheses for such an error, tracing linking pages, ... A fascinating puzzle but one that requires time and dedication :)

In general, manual manipulation is, of course, always possibe as the IA is not set up as a "trusted" repository in the sense of proving and documenting authenticity, as compared to regular archives. But then the likelyhood of such a manipulation happening on a random page would also be quite unlikely. And as pages are heavily cross-linked, it would be really hard to manipulate anything in this complex network without being inconsistent on some page or other.

In how much any evidence provided by the IA holds in court is a completely different question that I am not competent to answer. I have not traced any usage of such material in court cases, and specifically am not aware of any cases where such evidence was challenged. I guess, at least in our legal system, it would end up with one or the other side having to demonstrate convincing evidence for such an error being possible or not - and the willingness of IA to participate in this. (Or, in your case, as you have an email from them stating that the page existed, their willingness to state that this was wrong - not sure in how far they have to or would want to provide evidence for that.)

It is definitely an interesting situation, and I would be very interested to hear what evolves from it, both from a technical perspective (traceability of computation, trust in computing resources, data management) as well as from a sociological perspective (as such archives establish "truth", a view of the world, while we lack mechanisms to verify this - especially over longer periods in time). But I doubt that it will be easy to resolve this via the helpdesk of the IA, unless they themselves are interested in resolving this puzzle and would be willing to trace the tech ical aspects themselves to find an explanation that is plausible - which they should be interested in as well, I assume, given that they likely want to avoid any further such mistakes and figure out what other pages are affected to correct the error (while, I hope, documenting any such correction as would be good archival practice)

If there is any news, keep me informed - I'm curious to learn more!

best regards,

*******

I sent the Prof the link as suggested and highlighted the points that HKP requested. Interesting but again nothing definitive.

I have one query - does anyone have a record, from the 17 June, of the request and response headers for the mccann.htm page and the madeleine_02.jpg? I think the process of changing files from a 200 GET to a 302 GET causes the mementos to be removed. It would be very handy if anyone has got a screenshot of the full request/response header sets from either file before the changes took place.
thanks

Steve Marsden's WBM screenshot: The CEOP Home page for April 30, 2007 also refers to Missing Madeleine.