Perhaps you should take a look at Strawberry browser. Is on invite-basis only at the moment, not free either, because of AI, but the user interface appears to be well suited for (automating) research in combination with AI. Screenshots of the UI you can find here, as well as a complete description, a FAQ, some example animated Gif's, etc.
-Shades
Thanks. I'll take a look; I'm certainly going to be paying more attention to browsers generally.
doesn't disappoint in the same way.
-Shades
It may not seem like it, but I'm still very pleased about Horse. I'll admit that I now question whether the wonderful workflow I envisaged was somehow my own invention, but at the very least it was inspired by Horse. It should have been obvious - it was obvious! - but it had never occurred to me before.
Every now and then an interest in family history and genealogy takes hold of me. Because of the big gaps, I cannot remember the precise details of the work I have done or not done - especially failed searches. I know I should make detailed records, but I only ever do it incompletely, sometimes, and I'm unlikely to do better because it is a significant amount of drudgery for the remote prospect of a possible payoff in the distant future. This mostly automates the whole thing.
Off the top of my head, I recognise three types of search/research that seem to crop up in all domains.
- One is planned and systematic. Defined data gathering, statistical analysis, maths. What most people think of as typical science, though science uses all these types.
- The second is like a wood sculptor beachcombing for driftwood with potential. Probably one of the most common web activities. Despite being sometimes vilified as mindless collection, the items are often chosen as potential triggers for the imagination.
- And the third is like the hunt for Bigfoot; you know it's there, you're sure it's there, it's just very, very hard to find and most trails lead nowhere. And you just need a bigger collider.
And it's only the third type that I imagined Horse helping with. Not that it matters. Now that I have the idea, the Vivaldi method will work perfectly well enough.
Family history research is a funny beast. Most of the data you work with is in archives, which constantly acquire new data. So search results may change over time. And sometimes the site will only cough up the data if you search x way rather than y way because search algorithms
can be glitchy. The data is often mistranscribed and the original respondents were often misheard, or had limited knowledge, or downright lied. Does this record belong to that person?; does that person, or that record, truly belong in your family tree? Nothing is certain. Everything you 'learn' should have a probability estimate, but all you can do is write it down and keep some type of probability in your mind. You increase probabilities by triangulation. Preferably supported by copies of actual written records and not just transcriptions or someone else's assumptions. You constantly look for stuff to extend, but more often you look for supporting evidence, and even more often disconfirming evidence. A disappointing but useful outcome.
So covering old ground is the norm. And remembering the sites or the precise search terms is hard, but writing down every slight change in location or date range is tedious beyond belief. It had never occurred to me to store that information in a browser. Partly because the searching can seem haphazard - I tire of not finding Gobble Grimstone in Derby, exhaust my supply of variants, Gimstone, Gimston, Gimson, Jemson; and switch to Abel Turkey (probably Tukey oc) in Nottingham; and round and round in circles. Even worse I do find them, but with records that show they're not one of my ancestors and I have to start again. Partly because tabs always seemed haphazard themselves and using multiple browsers didn't help because I never saw them as central to my workflows. Seeing tabs as an outline was a real eye opener; I can shuffle and rename, even use emojis. Reorganise in the browser. And the power is amplified by moving the searches into a PKM notes app.
I appreciate the lure of AI in all of this; genealogy sites are using it directly in a limited way. But I'm also very wary. The data is already probabilistic. AI's internal logic must function around probabilities. I fear a tendency to be overly certain or even a willingness to invent a record that it has decided must exist. Even inventing an image of a Dog Latin birth register from an Irish parish wouldn't be beyond it. And I would have no way of weighing the AI's probabilities.
PS That was interesting. I looked at the Strawberry link. I'd never quite realised that Lattics would just open URLs in one of its own windows.
Recent Posts