WEBVTT 00:00:00.040 --> 00:00:05.900 Hello. In the Crown of Aragon 639 years ago, 00:00:06.559 --> 00:00:13.759 it was March 28, 1387. My name is Jonathan Seyfried. 00:00:14.160 --> 00:00:17.359 I'm a PhD candidate in history at the University 00:00:17.359 --> 00:00:21.820 of New Mexico, and this is the Historian's Notebook, 00:00:22.320 --> 00:00:26.980 a podcast about how history gets made. Season 00:00:26.980 --> 00:00:31.019 1 is titled Molt Cara Companyona. We're looking 00:00:31.019 --> 00:00:33.939 at a document from each day of the first year 00:00:33.939 --> 00:00:38.579 of the reign of King Joan I of Aragon and Queen 00:00:38.579 --> 00:00:42.600 Violant de Bar. Let's begin today's document. 00:01:06.349 --> 00:01:11.129 Today's document is from the Varia Register of 00:01:11.129 --> 00:01:16.569 Violant, and it is an intriguing one, to say 00:01:16.569 --> 00:01:22.750 the least. So the addressee is the name at the 00:01:22.750 --> 00:01:26.450 very beginning of the letter underneath La Reyna, 00:01:26.969 --> 00:01:32.790 and that name is Mossen Fransech. So somebody 00:01:32.790 --> 00:01:38.219 named Fransech is the recipient and is getting 00:01:38.219 --> 00:01:43.500 instructions now i feel like i've seen that first 00:01:43.500 --> 00:01:47.959 name before in conjunction with the officials 00:01:47.959 --> 00:01:51.620 maybe in the treasury somewhere oh i just wish 00:01:51.620 --> 00:01:56.000 i knew who this francesc was because that would 00:01:56.000 --> 00:02:00.579 help a lot i think now the content of the letter 00:02:00.579 --> 00:02:08.300 is in regard to certain properties and also a 00:02:08.300 --> 00:02:14.639 person named Bernat Vilella. Okay, a couple weird 00:02:14.639 --> 00:02:19.039 things about this document. In the first line, 00:02:21.039 --> 00:02:26.719 the the content is about... it's it's some kind 00:02:26.719 --> 00:02:31.680 of officials. They're called Scrivanies del Justicias. 00:02:32.319 --> 00:02:37.039 or Justiciat. So there's some kind of, like, 00:02:37.620 --> 00:02:41.219 well, I guess like writers of the judiciary, 00:02:42.180 --> 00:02:44.280 writers of the judiciary. So this could be like 00:02:44.280 --> 00:02:49.439 the clerks. It's, it's not any kind of office 00:02:49.439 --> 00:02:53.120 that we've seen before. It's not a notary, because 00:02:53.120 --> 00:02:56.539 I think Violant would have used notary. It's 00:02:56.539 --> 00:03:00.060 not a bailiff or a vaguer. These are These are 00:03:00.060 --> 00:03:04.000 scribes. This is the word for scribes. Scrivanes. 00:03:04.360 --> 00:03:09.219 So the writers of the judiciary in a certain 00:03:09.219 --> 00:03:14.439 via e loch. So the town and the surroundings. 00:03:15.639 --> 00:03:21.500 And then line two is where it gets juicy because 00:03:21.500 --> 00:03:30.599 the place names are Boria and Magallón. Violant 00:03:30.599 --> 00:03:34.219 doesn't refer back to this in the letter herself, 00:03:35.199 --> 00:03:40.020 but on March 10th and March 11th we read some 00:03:40.020 --> 00:03:44.539 documents about these two places. Those documents 00:03:44.539 --> 00:03:50.330 were the ones that Joan and Violant sent to 00:03:50.330 --> 00:03:55.129 the governor of Aragon with a severe reprimand. 00:03:55.210 --> 00:04:00.069 Why? Because they had heard that the governor 00:04:00.069 --> 00:04:05.569 of Aragon had been dragging his feet to confiscate 00:04:05.569 --> 00:04:08.349 on the order that they had placed. Joan and Violant 00:04:08.349 --> 00:04:11.949 had given him an order to confiscate properties 00:04:11.949 --> 00:04:17.779 from Bernat de Fortia, the brother of the former 00:04:17.779 --> 00:04:22.800 Queen Sibilla and so they wrote to him, Joan 00:04:22.800 --> 00:04:26.079 wrote on March 10th and Violant wrote on March 00:04:26.079 --> 00:04:29.500 11th saying you'd better carry out our orders 00:04:29.500 --> 00:04:32.100 or else there's going to be severe consequences 00:04:32.100 --> 00:04:36.839 and the tone was the most punishing that we've 00:04:36.839 --> 00:04:41.379 seen in the entire podcast so far. They were 00:04:41.379 --> 00:04:45.439 mad. and the places that were listed in that 00:04:45.439 --> 00:04:49.620 letter are the same places that now we see in 00:04:49.620 --> 00:04:55.120 this innocent looking administrative document 00:04:55.120 --> 00:05:00.720 being sent to a certain Mossen Francesc and what's 00:05:00.720 --> 00:05:04.339 happening with these places well there's been 00:05:04.339 --> 00:05:08.560 a kind of document i believe it would it's what's 00:05:08.560 --> 00:05:13.930 referred to as the donades something like donation 00:05:13.930 --> 00:05:18.430 or kind of I'm not quite sure how to translate 00:05:18.430 --> 00:05:22.149 that in this legal sense but the writers of the 00:05:22.149 --> 00:05:28.029 judiciary have come up with some kind of documentation 00:05:28.029 --> 00:05:31.850 or authorization for these two locations and 00:05:31.850 --> 00:05:35.569 guess who is going to be receiving them it's 00:05:35.569 --> 00:05:37.829 someone I don't know we haven't heard of before 00:05:37.829 --> 00:05:41.129 so you wouldn't really be able to guess but it's 00:05:41.129 --> 00:05:45.430 someone who's closely connected to Joan and Violant 00:05:45.430 --> 00:05:48.430 and how do we know that? It's because on line 00:05:48.430 --> 00:05:53.870 two continuing on we have the name Bernat Vilella 00:05:54.279 --> 00:05:59.939 and then after that name we have De casa de Nostra 00:05:59.939 --> 00:06:05.740 Molt car Primogenit going on to line three lo dalfi 00:06:05.740 --> 00:06:12.439 de Girona so this is somebody Bernat Vilella who 00:06:12.439 --> 00:06:20.839 is of the house of our most dear firstborn inheritor 00:06:20.839 --> 00:06:28.240 to the throne the Dauphin de Girona that is Jaume 00:06:28.240 --> 00:06:32.339 their three -year -old son so there's someone 00:06:32.339 --> 00:06:37.089 that is an administrator for the affairs of their 00:06:37.089 --> 00:06:43.269 son and that person is Bernat Vilella and he now 00:06:43.269 --> 00:06:47.589 is going to be the recipient of these properties 00:06:47.589 --> 00:06:49.449 if I'm reading this letter correctly I think 00:06:49.449 --> 00:06:54.529 I am so that's what this whole letter is about 00:06:54.529 --> 00:06:57.269 basically saying all of the things that are related 00:06:57.269 --> 00:07:02.360 to these properties should now be handled by 00:07:02.360 --> 00:07:08.420 Bernat Vilella and that's the whole thing basically 00:07:08.420 --> 00:07:12.480 and and at the end it's uh we've got a certificants 00:07:12.480 --> 00:07:18.839 nos so that Violant will be informed when Mossen 00:07:18.839 --> 00:07:24.550 Francesc has processed this order to get these 00:07:24.550 --> 00:07:30.790 properties into the control of Bernat Vilella. 00:07:31.050 --> 00:07:37.709 So it seems that the governor of Aragon did respond 00:07:37.709 --> 00:07:43.329 to those very harsh letters that Joan and Violant 00:07:43.329 --> 00:07:48.329 sent on March 10th and March 11th and that by 00:07:48.329 --> 00:07:52.709 March 28th those properties are now available 00:07:52.709 --> 00:07:58.290 to be given to somebody else. So interesting. 00:07:58.569 --> 00:08:02.350 to note that because there's nothing in the document 00:08:02.350 --> 00:08:05.850 itself that's like celebrating the successful 00:08:05.850 --> 00:08:09.389 confiscation of these properties from the brother 00:08:09.389 --> 00:08:14.290 of their nemesis, Sibilla. There's nothing in 00:08:14.290 --> 00:08:17.509 the document itself indicates that, but if you 00:08:17.509 --> 00:08:20.930 put it together, if you corroborate one of these 00:08:20.930 --> 00:08:25.209 historical thinking practices corroboration if 00:08:25.209 --> 00:08:28.170 you corroborate it with something we saw before 00:08:28.170 --> 00:08:33.590 then we have an inference to make and the story 00:08:33.590 --> 00:08:39.690 gets continued with an earlier document so that's 00:08:39.690 --> 00:08:43.629 the ideal thing that a historian is trying to 00:08:43.629 --> 00:08:46.970 do when working with an archive that doesn't 00:08:46.970 --> 00:08:51.679 really give you the answers to the questions 00:08:51.679 --> 00:08:55.840 you find most interesting sometimes. Other times 00:08:55.840 --> 00:08:59.519 it does, but I just had a ton of questions about, 00:08:59.740 --> 00:09:02.159 well, you know, what was going to happen next 00:09:02.159 --> 00:09:06.080 with these properties, and I wanted to know why 00:09:06.080 --> 00:09:08.879 the governor of Aragon had been dragging his 00:09:08.879 --> 00:09:10.620 feet in the first place. We don't get anything 00:09:10.620 --> 00:09:15.120 about that, and we still don't, but we do know 00:09:15.120 --> 00:09:21.200 that Joan and Violant could, if they were 00:09:21.200 --> 00:09:24.700 threatening enough they could get what they wanted 00:09:24.700 --> 00:09:27.460 from their subordinates which is really important 00:09:27.460 --> 00:09:31.679 because it would it would be a really dramatic 00:09:31.679 --> 00:09:35.340 thing and important to know if the subordinates 00:09:35.759 --> 00:09:40.639 would continue to try to drag their feet after 00:09:40.639 --> 00:09:43.960 receiving a threatening letter like the one that 00:09:43.960 --> 00:09:49.059 came from Joan on March 10th. That would be really 00:09:49.059 --> 00:09:52.200 something if the governor of Aragon had continued 00:09:52.200 --> 00:09:57.960 to circumvent the will of the monarchs, but We 00:09:57.960 --> 00:10:01.860 see that the exercise of power, to the contrary, 00:10:02.259 --> 00:10:06.080 has been successfully carried out. That Joan 00:10:06.080 --> 00:10:10.379 and Violant are building up their ability to 00:10:10.379 --> 00:10:13.539 catch people who are circumventing their will 00:10:13.539 --> 00:10:19.720 and that they are able to make their will followed, 00:10:20.360 --> 00:10:23.940 at least in this particular instance. so we can 00:10:23.940 --> 00:10:29.179 chalk this up as a victory for the power and 00:10:29.179 --> 00:10:34.200 exercise of power of the King and Queen at the 00:10:34.200 --> 00:10:38.460 beginning of their reign Alright, that's what 00:10:38.460 --> 00:10:42.899 I got for this particular document I want to 00:10:42.899 --> 00:10:46.460 give you an update about what I am doing with 00:10:46.460 --> 00:10:54.299 AI usage kind of exciting I have started to explore 00:10:54.299 --> 00:11:01.779 the world of agentic AI, which means when you 00:11:01.779 --> 00:11:09.139 turn over a lot of tasks to an AI that you've 00:11:09.139 --> 00:11:15.179 given permission to follow instructions and act 00:11:15.179 --> 00:11:22.100 on your own machine. So I took an old laptop, 00:11:23.620 --> 00:11:26.820 and it's speedy enough for the purposes at hand 00:11:26.820 --> 00:11:30.019 here. And that's because a lot of the processing 00:11:30.019 --> 00:11:35.899 is done by the AIs. you know data center and 00:11:35.899 --> 00:11:40.779 using a lot of a lot of energy and Water in order 00:11:40.779 --> 00:11:44.440 to do that. So that's distressing ethically again, 00:11:44.700 --> 00:11:46.460 you know, I've talked about the ethics of AI 00:11:46.460 --> 00:11:50.259 before it's good to always mention that and be 00:11:50.259 --> 00:11:55.500 aware of it and not to use it frivolously because 00:11:55.500 --> 00:11:59.779 it is costly from an environmental point of view 00:11:59.779 --> 00:12:03.529 and So we don't want to forget that and at the 00:12:03.529 --> 00:12:07.470 same time these tools are extremely powerful 00:12:07.470 --> 00:12:11.190 and have tremendous potential for unlocking a 00:12:11.190 --> 00:12:16.830 lot of historical information and providing historians 00:12:16.830 --> 00:12:21.730 with tremendous new capabilities for understanding 00:12:21.730 --> 00:12:26.169 the past and creating narratives of the past 00:12:26.169 --> 00:12:29.250 and we should explore that because the more that 00:12:29.250 --> 00:12:34.500 we can innovate in historiography I think the 00:12:34.500 --> 00:12:43.159 better for society. So using a lot of AI for 00:12:43.159 --> 00:12:47.720 this podcast is something that I think is worth 00:12:47.720 --> 00:12:54.419 the cost and also the ethical problems that those 00:12:54.419 --> 00:12:58.759 costs are. All right let's get back to what I'm 00:12:58.759 --> 00:13:03.240 exploring and doing now. So on this other computer 00:13:03.240 --> 00:13:06.899 that does not have access to sensitive things 00:13:06.899 --> 00:13:10.360 like my tax returns and stuff on my my normal 00:13:10.360 --> 00:13:14.500 computer this computer has been wiped i have 00:13:14.500 --> 00:13:19.620 downloaded openclaw you might have heard of openclaw 00:13:19.620 --> 00:13:24.279 it's really the most popular ai agent these days 00:13:24.279 --> 00:13:28.600 and you know this is march of well the end of 00:13:28.600 --> 00:13:31.559 march of 2026 and so of course that could change 00:13:31.559 --> 00:13:36.059 what open claw does is it helps to do things 00:13:36.059 --> 00:13:38.500 on your computer but it always is going to be 00:13:38.500 --> 00:13:42.899 working with some other ai providers so you get 00:13:42.899 --> 00:13:45.600 to choose when you download open claw whether 00:13:45.600 --> 00:13:49.620 it's going to be working primarily with clawed 00:13:49.620 --> 00:13:54.710 or with uh chat gpt So you make that choice I 00:13:54.710 --> 00:14:03.490 chose Claude and so I have the images from the 00:14:03.490 --> 00:14:09.350 registers on my this this other computer and 00:14:09.350 --> 00:14:13.049 I Tested out open claw to see if it would be 00:14:13.049 --> 00:14:17.909 able to do the process that I do so for the last 00:14:17.909 --> 00:14:21.200 several episodes what I've done is I've taken 00:14:21.200 --> 00:14:25.720 the image of the document, and I have had Gemini 00:14:25.720 --> 00:14:30.559 do a transcription, gotten okay results. Not 00:14:30.559 --> 00:14:33.200 really, I wouldn't call it usable, well sometimes 00:14:33.200 --> 00:14:35.639 it is, but it's on the borderline of usable and 00:14:35.639 --> 00:14:40.179 not usable. And then I have Claude do an initial 00:14:40.179 --> 00:14:44.200 transcription, and then I have the two get reconciled 00:14:44.200 --> 00:14:48.080 by Claude. So the output from Gemini helps Claude 00:14:48.080 --> 00:14:53.159 to improve its transcription. then Claude does 00:14:53.159 --> 00:14:57.159 that reconciliation and I've been giving it feedback 00:14:57.159 --> 00:15:01.759 after that reconciliation and then it does a 00:15:01.759 --> 00:15:05.600 final transcription based on my feedback and 00:15:05.600 --> 00:15:10.600 then a provisional translation. That's been my 00:15:10.600 --> 00:15:16.480 process and I can report to you that OpenClaw 00:15:16.480 --> 00:15:20.279 is able to do all of that with one exception. 00:15:20.720 --> 00:15:25.320 The part where I give Claude my feedback before 00:15:25.320 --> 00:15:29.480 it does that third stage of a final transcription, 00:15:30.000 --> 00:15:35.679 that is not something that is possible when OpenClaw 00:15:35.679 --> 00:15:39.759 does this. But OpenClaw has been able to do, 00:15:39.759 --> 00:15:43.639 and I've done it for one register, it's able 00:15:43.639 --> 00:15:50.450 to do over 250 images of all the pages in the 00:15:50.450 --> 00:15:53.529 register. Okay, I think it was like 297 total. 00:15:54.549 --> 00:15:57.570 Yeah, I think that's right. So 297 images, it 00:15:57.570 --> 00:16:02.669 was able to do that on its own. So I now have 00:16:02.669 --> 00:16:13.379 a transcription of 297 pages from register 1819 00:16:13.379 --> 00:16:16.320 that was the first one I did a test for and it 00:16:16.320 --> 00:16:20.059 took open clot working with Gemini and Claude 00:16:20.059 --> 00:16:23.919 it took it about eight hours to do that and it 00:16:23.919 --> 00:16:28.120 also had a cost because when you use the when 00:16:28.120 --> 00:16:32.240 you use the AI chat bots I don't even know if 00:16:32.240 --> 00:16:34.330 you can Yeah, I guess they're still chatbots. 00:16:34.389 --> 00:16:37.590 When you use the LLMs this way, the companies 00:16:37.590 --> 00:16:42.090 will charge you money. So it's not going to be 00:16:42.090 --> 00:16:45.610 economically feasible for me to do this for every 00:16:45.610 --> 00:16:54.370 register that contains content for 1387, but 00:16:54.370 --> 00:16:59.210 I'm going to do it for a few. And then I'm able 00:16:59.210 --> 00:17:06.430 to do a keyword search. So, for example, we had 00:17:06.430 --> 00:17:12.349 a document a few episodes ago in which Violant 00:17:12.349 --> 00:17:17.390 wrote to her father about how Joan had recovered 00:17:17.390 --> 00:17:22.750 from his illness. And that was in Register 1819. 00:17:23.490 --> 00:17:29.750 And I was able to use full text searching from 00:17:29.750 --> 00:17:33.309 what OpenClaw generated in order to find that 00:17:33.309 --> 00:17:39.029 letter when I asked it to search for transcripts 00:17:39.029 --> 00:17:43.930 that were about illness and had the word malaltia. 00:17:44.710 --> 00:17:49.250 So we are now at the point where we've got I 00:17:49.250 --> 00:17:53.490 would call it a rudimentary full text searching 00:17:53.490 --> 00:17:58.630 capability for late 14th century gothic secretarial 00:17:58.630 --> 00:18:01.670 hand documents from the archive of the crown 00:18:01.670 --> 00:18:06.349 of aragon so that's a pretty awesome result it 00:18:06.349 --> 00:18:09.890 is something i thought i would get through transkribus 00:18:09.890 --> 00:18:14.250 but it's been it's been rough going with transkribus 00:18:14.250 --> 00:18:18.509 i've done 60 transcriptions of documents by hand 00:18:18.509 --> 00:18:22.039 for Transkribus in order to give it some ground 00:18:22.039 --> 00:18:25.619 truth to train on, and I gotta tell you that 00:18:25.619 --> 00:18:28.859 the process I have with Gemini and Claude is 00:18:28.859 --> 00:18:31.839 better than what Transkribus is able to produce 00:18:31.839 --> 00:18:35.839 at this time. So I'm going to update the Season 00:18:35.839 --> 00:18:41.240 1 FAQs about the OpenClaw process on that section 00:18:41.240 --> 00:18:44.259 that I have about AI usage. And you'll start 00:18:44.259 --> 00:18:47.339 to see some different notes about AI usage on 00:18:47.339 --> 00:18:51.819 the web pages for documents. And I can't use 00:18:51.819 --> 00:18:55.940 this all the time, every time, because it costs 00:18:55.940 --> 00:19:01.339 about $50. to do a single register so I mean 00:19:01.339 --> 00:19:04.000 yeah I cannot do there's and there's probably 00:19:04.000 --> 00:19:08.039 about 40 registers that have information from 00:19:08.039 --> 00:19:12.059 this year in Joan and Violant's reign 1387 so 00:19:12.059 --> 00:19:14.740 not going to be able to not be able to do that 00:19:14.740 --> 00:19:19.740 but it is a proof of concept for sure and it'll 00:19:19.740 --> 00:19:22.220 help me with the registers that seem to have 00:19:22.220 --> 00:19:25.420 the most interesting and relevant content for 00:19:25.420 --> 00:19:29.380 my research question. So that's the update on 00:19:29.380 --> 00:19:32.640 AI usage. It's getting interesting. I mean, we're 00:19:32.640 --> 00:19:36.680 still at the beginning of 2026 and I am just 00:19:36.680 --> 00:19:41.859 astonished at how far AI has come in just the 00:19:41.859 --> 00:19:45.200 last three months. And if you then turn the clock 00:19:45.200 --> 00:19:48.500 back even further and think about where AI was 00:19:48.500 --> 00:19:55.029 in say October of 2025, this is a pace of improvement 00:19:55.029 --> 00:19:59.630 that is really, you know, really something. Whew. 00:20:00.490 --> 00:20:04.809 Okay. Well, thanks for listening to this episode 00:20:04.809 --> 00:20:08.950 of the Historian's Notebook, Season 1, Molt Cara 00:20:08.950 --> 00:20:12.269 Companyona. If you are leaving with more questions 00:20:12.269 --> 00:20:15.730 than you arrived with, I have done my job. Remember, 00:20:16.089 --> 00:20:20.569 the motto of the Historian's Notebook is dissatisfaction 00:20:20.569 --> 00:20:23.819 guaranteed. Visit the website to see an image 00:20:23.819 --> 00:20:26.700 of today's document and additional show notes, 00:20:27.240 --> 00:20:30.220 and listen again tomorrow to hear about the next 00:20:30.220 --> 00:20:33.839 day in the first year of the reign of King Joan 00:20:33.839 --> 00:20:38.660 I of Aragon and Queen Violant de Bar. In the 00:20:38.660 --> 00:20:40.079 meantime, take care!