Dear all, 

Like Wikipedia, ChatGPT depends on the quality of its sources. When you ask it historical questions in some sense it is just autogenerating a summary of existing literature, as Wikipedians do by hand. But it can get confused, and my own results were distinctly mixed. (Note that all this was a couple of weeks ago, so ChatGPT may have also changed a bit). 

First I tried asking ChatGPT to list it’s sources for a question about whether people used fill-in forms to author pages on early wikis, with some but not complete success – see below. I was feeling pretty good about it as an automated research assistant. 

But THEN, it gave me two contradictory answers when I asked a question twice but in different words about fill-in forms for building pages in GeoCities. First it said GeoCities DID use fill-in forms for authoring directly on the site. Then it said it did NOT, that users had to create pages offline and upload them. 

Both answers are partly true, in the sense that GeoCities had at least three online page-builders and you could also upload pages created offline. But... that reality is exactly what an accurate answer should have described! 

I still find ChatGPT amazing for quickly summarizing info and for pointing out directions I might not have thought of, but EVERYTHING it says needs fact checking. 

Best, Marc


Wiki example, asking ChatGPT to summarize sources: 

Q: Did early wikis use fill-in forms for authoring? please cite sources

A: Yes, many early wikis used fill-in forms for authoring. The first wiki, WikiWikiWeb, which was created by Ward Cunningham in 1995, used a fill-in form for editing and creating pages. According to the WikiWikiWeb website, the form included fields for the page title, the author's name, and the page content. Additionally, Ward Cunningham in his own words, discussing the early days of Wiki: "..I thought I would make it easy to create new pages by having a form that could be filled in right on the screen.."

Source:

Q: How about for fill-in forms in other popular early wikis?

A: Many early wikis, including popular ones, used fill-in forms for authoring pages.

For example,

This is widely adopted method as it is simple, intuitive and accessible for most people and it allows for version tracking and simple page linking.

You can find more information about these and other early wikis on their respective websites and in publications that discuss the history of wikis.

Sources:

Marc Weber (he/him)
Curatorial Director, Internet History Program
Computer History Museum, 1401 N Shoreline Blvd., Mountain View CA 94043
marc@webhistory.org  |  +1 415 282 6868
computerhistory.org/nethistory  |  Co-founder, Web History Center and Project

On Feb 3, 2023, at 13:26, Ceruzzi, Paul via Members <members@lists.sigcis.org> wrote:

Thanks, Jim. 
I plan to check it out.

On a somewhat related topic, I tried out ChatGPT (so you don't have to!), and I asked it who was the source of the quote that only 5 computers would serve the world's needs. A topic, by the way, that was discussed extensively in past issues of theAnnals​. It came up with:

Thomas J. Watson, the president of IBM, is often quoted as having said in 1943 that "I think there is a world market for maybe five computers." This statement is often cited as an example of a bold prediction that was ultimately proven to be incorrect, as the number of computers in use today is vastly greater than five.

Strictly speaking, it is not wrong in that Watson is often quoted, but by whom? And who really was the source of the quote? Typical weasel by ChatGPT to use the passive voice and avoid answering the question.

I also asked it for a list of pioneers of space science and technology, and it came up with, among the usual pioneers,  Arnold Sommerfeld. ??? I wonder if that was an example of the "Matthew Effect": Wikipedia gives Sommerfeld credit for so much of modern physics, why not credit him with the development of rocket science as well? He was a great physicist but does not belong on that list.

Its replies are grammatically correct and free of spelling or punctuation errors, which is impressive, at least to me.  And it got other things right. I guess it draws heavily on Wikipedia, but obviously not the Annals. Why not just go to Wikipedia directly?  I found it pretty easy to break, which makes me think this may be a modern version of Joe Weizenbaum's ELIZA.

Best,

Paul Ceruzzi
From: Members <members-bounces@lists.sigcis.org> on behalf of James Cortada via Members <members@lists.sigcis.org>
Sent: Friday, February 3, 2023 12:23 PM
To: sigcis <members@sigcis.org>
Subject: [SIGCIS-Members] A New Book
 
External Email - Exercise Caution
A book I have written has just been published and thought it might be of interest to this community: Birth of Modern Facts.  I explore how information has changed over the past 150-200 years by looking at these transformations through the lens of disciplines (e.g., computer science, library science, physics, economics, etc.) and their ecosystems.  Every chapter includes discussions about the role of computing, too.  The publisher (R&L) has offered a discount to individuals, for which I attach its coupon.  Should you like to discuss my ideas in the book I am all ears, because I am still exploring this broad theme and one book cannot settle the issue in my mind.

Regards,

Jim

-- 
James W. Cortada
Senior Research Fellow
Charles Babbage Institute
University of Minnesota
608-274-6382
_______________________________________________
This email is relayed from members at sigcis.org, the email discussion list of SHOT SIGCIS. Opinions expressed here are those of the member posting and are not reviewed, edited, or endorsed by SIGCIS. The list archives are at http://lists.sigcis.org/pipermail/members-sigcis.org/ and you can change your subscription options at http://lists.sigcis.org/listinfo.cgi/members-sigcis.org