This week’s readings, specifically those on crowdsourcing and Transkribus, reminded me of a project I worked on at the Niagara Falls Historical Museum. They have a huge project for volunteers that involves transcribing 59 handwritten ledgers from the 1831 Niagara Falls Museum founded by Thomas Barnett. The ledgers date back to 1838 and span more than 150 years, so there is a lot of work to be done. Currently, the museum sends the digitized files out to their volunteers and allows them to work on transcribing whenever they have time to spare. While reading through the crowdsourcing guide for libraries, I couldn’t help but think how perfect crowdsourcing this project would be. Instead of having to send out the digitized files and check in with their volunteers, they could publicize the pictures and create a platform where anyone could go on and transcribe these ledgers. Some of these ledgers have some very famous signatures (e.g. Sir John A. MacDonald, Abraham Lincoln, General Ulysses S. Grant, and P.T. Barnum), which could be a way of getting the attention of more people. Likewise, there is also the chance that people have family members who would have visited Niagara Falls and signed one of these ledgers, which could also attract more people to help. Right now, the pages that have been transcribed can be searched online.
I also tried to look into how exactly Transkribus works, and whether it would be applicable for this project. While this project would be complicated for Transkribus (since it has many different handwriting styles), I wondered whether what someone else has “taught” the program (say, someone who has already digitized and transcribed documents written by Abraham Lincoln) could be used by the museum to have his signature transcribed automatically. I couldn’t tell for sure, but what I read made it sound like each person has to train Transkribus to recognize the documents that they are using. It did make me wonder about a universal version of this program; perhaps a mix of crowdsourcing, in that anybody can upload and transcribe any sort of document they want (perhaps even specific words or phrases in their own handwriting), along with the institutional uses that Transkribus has, in order to teach an AI program that can they learn and be used by the masses (instead of keeping the training specific to the person, like Transkribus).
I also found the suggestion for libraries to digitize their images and allow them to be catalogued and described by the public to be very interesting (and also very applicable for the Niagara Falls History Museum). Having gone through the process of digitizing, cataloging, and describing artifacts, I can comment on the large number of hours it takes. Being able to take some of those steps away from the institutional workers, and allow the public to help, allows for employees to spend their time on other projects while also engaging the public and giving them an easy way to volunteer their time. As a more general note after reading through the crowdsourcing suggestions, I did find myself wondering how everything would be moderated, since the article doesn’t really touch on this. Just as with Wikipedia, I’m sure there would be instances of people messing around with crowdsourcing projects (especially in an online setting). How would institutions like libraries and museums make sure that only those legitimately wanting to help are allowed to volunteer via crowdsourcing? I’m sure staff could go over what would be uploaded, but that might take almost as much time as doing the crowdsourced work.