January 13, 2004

Our Nightmare Is Over

Well, thank God that nightmare is over. It's depressing: we're clearly moving toward registered comments only, for all the standard reasons...

A new year, and time to start a new database and a new archive directory. So the last few days of postings can now be found here.

Posted by DeLong at January 13, 2004 02:34 PM | TrackBack



Can you tell us a bit more about the forces of unrighteousness behind the 'comment spam'? Was it the Krugman stalkers? And what exactly did they do?

Posted by: Will on January 13, 2004 03:11 PM


Professor, I would be happy to register. I wonder what the privacy concerns would be. I assume your regular readers trust that you would not abuse the registry. Most of the regular commenters give an address when they post anyway. While that might be of some concern for some people, I expect that most of us know how to handle several e-mail addresses, and to change them when we need help with weird stuff. I wonder what other concerns there might be.

Posted by: Masaccio on January 13, 2004 03:46 PM


A shame it's having to come to this, but what alternative is there, really? Before making that final leap, however, why not try James Seng's Bayesian filtering plugin


and see how it works? It seems likely to me to be a lot more scalable in the long run than either IP blocking or dealing with individual registrations.

Posted by: Abiola Lapite on January 13, 2004 04:04 PM


You missed an "h" in nielsenhayden link, so it doesn't work. This one should:


Posted by: bulent on January 13, 2004 04:06 PM


I'm interested in the 'what exactly they did' answer too as I have multiple MT-based sites and want to be prepared.

Posted by: JB on January 13, 2004 05:13 PM


It must be like being sprayed ink by octopi !

Or like a single big info-organism, with stingers...

Posted by: Lee A. on January 13, 2004 06:25 PM


Masaccio: The privacy concerns are whatever you want to make them. These websites are out there in the public, and search engines like google and others can be used to catalog and store them. All information in the pages that the search engine traverses is exposed, and while google (presumably?) does not include tag values, including the posters' email addresses, in the cataloging, others may. Even if the email addresses or personal information are not included on the comments pages in the registration scheme, they are probably stored somewhere on the server or the database engine, which may be open to hacker attacks. Generally information safety on computers should not be trusted a lot. You all know how credit card records, customer lists, etc. have been compromised.

It is known that even today employers, insurances, loan offices etc. do background checks on individuals, and use stereotyped categories ("profiling") to judge them. For example, it is at least generally speculated that (auto?) insurers use features like credit scores or lack of credit card ownership in their risk models. Who knows, maybe they discover (or hypothesize) correlations between posting on left-leaning forums and things that are relevant to their modeling? (This example is only half tongue-in-cheek.)

The problem with such profiling and stereotyping approaches is not just discrimination of certain groups (which may be good or bad depending on the context), but that the discrimination is done by computer, and based on more often than not incorrect or irrelevant data, e.g. by processing errors or sloppiness in compiling the databases. In other words, computer don't improve your judgement, they only amplify it. There is also a possible positive feedback -- one insurer (let's say) rejects or labels you based on some original incorrect information, the next one based on this rejection (if the first guy rejected you, there must be something wrong with you). Even if people intend well, you wouldn't believe how many think the computer is "always right".

To give you a somewhat unrelated example, it happened to me that a collection agency called me up trying to reach somebody who previously lived at my place. The representative claimed that we were listed in their database as friends or family, which was not true. What is true, however, is that our phone number is listed on this address, whereas the number of our landlord is not. There are multiple scenarios what may have happened -- (1) the representative made up the friends thing, (2) the friends thing is in their database, (3) our landlord used to be in the database but got his number unlisted, and they looked it up again. The point here is that the information they stated was incorrect, and this kind of guesswork probably happens more often than you think.

The only effective recourse is to minimize having your identifying information in databases.

Posted by: cm on January 13, 2004 08:19 PM


This is such a treaure.
What a shame to be spammed.

Posted by: anne on January 14, 2004 09:49 AM


Post a comment