I will very briefly reply for the benefit of the list, although it is a
small matter in my opinion. No doubt the DSpace version of the button
does need some spam protection. I made it quite clear that the email
address was never acquired by the spammer, and that merely a web form
was spammed. This is pretty simple for a programmer to achieve where a
tool is undefended. Compare for instance how the OpenFRBR web tool has
been spammed by an automated crawler script. I'm sure the DSpace
developers will address the issue in future.
The point was raised as an illustrative one only. The spam was entirely
secondary to the issue at hand, which is the usefulness of the button.
We did not use the button for other items because the institutional view
at the time was that it was of uncertain legality without a publisher's
explicit permission. As I no longer work for my former employer, I will
not comment on their institutional liabilities except to say that they
are in the company of many other universities in preferring caution in
an untested area of law.
Arthur Sale wrote:
> Can you explain then how the spammer got the address if you never revealed
> it? Are you suggesting that the spammer submitted genuine requests and then
> manually harvested email addresses from the fulfilled responses with the
> paper? This is stretching credulity too far.
> Or are you suggesting that a form page following the Request-a-copy button
> was spammed with multiple spurious messages in a text box? This is indeed
> more difficult to counter, but not very much. A competent software engineer
> can filter out most such spam messages by textual analysis. The spam problem
> is not new. You could even have replaced the box by a checkbox affirmation,
> which might have been better, and have removed all spamming incentive. If
> this was the case, at least the author's email address was not exposed to
> spammers, as little as that helps these days! I do understand however that
> you exposed only one item to the button, and thus did not spend much effort
> on implementing the button.
> Arthur Sale
> Emeritus Professor of Computer Science
> University of Tasmania
> -----Original Message-----
> From: Talat Chaudhri [mailto:[log in to unmask]]
> Sent: Monday, 30 March 2009 7:22 PM
> To: Arthur Sale
> Cc: [log in to unmask]
> Subject: Re: Central 'Request a copy' address?
> I should clarify, Arthur. We of course did not expose the author's email
> address to spam. Web forms can also be spammed on an automated basis,
> which is what happened here. This is considerably more difficult to
> prevent. However, the email address was never revealed.
> Arthur Sale wrote:
>> It is important in implementing the button to NOT expose the author's
>> address to crawlers through the html (eg by a mailto link). The button
>> should go to a script that generates the email to the author or authors.
>> Crawlers cannot follow the script.
>> This duty is indeed a responsibility of repositories generally.
>> We have had no reported spam at all. I don't see how it could arise anyway
>> given a safe implementation.
>> My impression is that the button is not used very often, but every time I
>> have used it myself, it results in a delivery of the document. I don't
>> hard statistics on usage.
>> Arthur Sale
>> University of Tasmania
>> -----Original Message-----
>> From: Repositories discussion list
> [mailto:[log in to unmask]]
>> On Behalf Of Talat Chaudhri
>> Sent: Saturday, 28 March 2009 12:30 AM
>> To: [log in to unmask]
>> Subject: Re: Central 'Request a copy' address?
>> I should add perhaps that the idea of OpenID, of course, would be that
>> people can retain the same OpenID for multiple purposes and update the
>> attached contact details as they move about. However, the issue at hand
>> is whether people really do this, or whether they will in future as
>> OpenID develops. Lots of services offer OpenIDs as a side benefit but
>> people aren't always aware of it, and many then abandon them when they
>> stop using the service. Seems like a work in progress in many ways, but
>> in principle it could help address the issue of keeping contact details
>> Talat Chaudhri wrote:
>>> I won't reiterate the points made in earlier debates: as Charles says
>>> succinctly in his reply, the short answer is no.
>>> However, I will share my experience at Aberystwyth in my previous
>>> role. We used the button where the publisher agreed to this, for one
>>> item. It received a great deal of spam and the author complained. We
>>> therefore switched the item from his LDAP user to the administrator by
>>> changing the database entry, so that we received the email instead and
>>> could monitor it. In two years or so of using the button on this item,
>>> we only ever received one genuine request that we forwarded to the
>>> author. I have no idea whether he then approved the request, or how
>>> serious a request it was, as we had no further response. I may add
>>> that the author had other papers in the repository that extremely high
>>> monthly access statistics, by way of comparison.
>>> The problem would be greater where a member of staff had left the
>>> institution. The the email contact address, however the button may
>>> work on a particular platform in practice (DSpace using LDAP accounts
>>> at present), would then be redundant, as would the button except in
>>> the rare case that details would be updated for a defunct account. One
>>> possible answer is to rely on OpenID but there is still no guarantee
>>> that a particular OpenID would remain valid, any more than an email
>>> address. Overall, it brings the practical usefulness of the button
>>> into considerable doubt.
>>> I gather from Les' comment that the Eprints button works rather better
>>> because it can be associated with multiple email addresses and does
>>> not rely on the LDAP mechanism, but it is quite possible that several
>>> authors might move on or change email addresses, so the problem is
>>> merely reduced.
>>> Does anyone have statistics for the actual use of the button, i.e.
>>> number of requests as a percentage of total access, number of requests
>>> agreed and refused by the author etc, for a particular repository? It
>>> would be very interesting to see whether our experience was unusual or
>>> typical. I would not like to advocate the button until it is shown
>>> that people really use it, laying aside the possible legal intricacies
>>> mentioned on earlier occasions.
>>> J.W.T.Smith wrote:
>>>> In EPrints, when there is a contact address for a repository item, an
>>>> external user sees a 'request a copy' button.
>>>> I was thinking of adding a generic 'request' address to all the items
>>>> that have no contact address so requests for these items would come
>>>> to a central service. If I have understood Charles Oppenheim's advice
>>>> on Copyright we could supply a copy of the paper to the requestor
>>>> free of charge without infringing Copyright (assuming they say it is
>>>> for private non-commercial use).
>>>> Has anyone done this (or similar)? Is it Copyright OK?
>>>> John Smith,
>>>> KAR (Kent Academic Repository) Admin.
Dr Talat Chaudhri
UKOLN, University of Bath, Bath BA2 7AY, Great Britain
Telephone: +44 (0)1225 385105 Fax: +44 (0)1225 386838
E-mail: [log in to unmask] Skype: talat.chaudhri