User talk:Sean.hoyland

Khirbet 'Ein Karzaliyah (Arabic: خربة عين كرزلية), Jordan Valley: December 2013 - January 2014

Id'eis (Arabic: ادعيس), Jordan Valley: May 2014

This user talk page might be watched by friendly talk page stalkers which means that someone other than me might reply to your query. Their input is welcome and their help with messages that I cannot reply to quickly is appreciated.

Archives

This page has archives. Sections older than 5 days may be automatically archived by .

Question (2)

Hi Sean,

I asked you recently how I can search for certain text in precious revisions of a page and I wasn't able to get it to work.

I'm currently trying to find if the name "Albanese" ever appeared in the article weaponization of antisemitism. Could you help me with advice on how to do that? I just tried WikiBlame but I don't it was working correctly.

Thank you, IOHANNVSVERVS (talk) 01:14, 25 March 2025 (UTC)[reply]

In a reference. Sean.hoyland (talk) 07:46, 25 March 2025 (UTC)[reply]

Thank you. IOHANNVSVERVS (talk) 11:45, 25 March 2025 (UTC)[reply]

Is there an "objective" way to measure article bias?

Hey, this might finally out me as a luddite, but with what looks like strengthening consensus that Israel is some sort of biased against Israel, I was wondering if there was some clever technical solution to find out if the 'tone' of articles and of high-quality sources are matching. I know that there are some approaches to use AI to analyse sentiment in text, but thought that you might have a useful idea beyond "look at what the Newspapers of record are saying, and then look at Hamas", which seems clear, but also not particularly scientific. FortunateSons (talk) 08:23, 1 April 2025 (UTC)[reply]

Hello there. The technical background to measuring bias is one several black holes out there that I have been fairly carefully avoiding learning too much about. There is a risk of getting trapped behind some kind of attention event horizon for weeks or months. Fortunately, I've managed to maintain my ignorance level so that my thoughts about it are probably "not even wrong". For example, I don't really even understand how it is possible in principle, let alone practice, because of the way Wikipedia treats neutrality as a function of source sampling rather than with reference to some value system or constitution-like thing (although there is a bit of that too). A large number of sources, perhaps the majority of sources, that describe something may not be sampled in practice, so the question 'biased with respect to what reference frame?' is a point of confusion for me. Do you need to look at nearly all sources to properly measure bias in a Wiki-context? I have no idea. BUT...the people building the big foundation LLM models have a similar problem. They need to be able to measure bias in the output text, so there might be something useful in the vast number of papers about that subject. However, they have the advantage of being able to compare things to reference sets like social norms. Sean.hoyland (talk) 09:20, 1 April 2025 (UTC)[reply]

Thank you very much for the detailed response; if you ever do get sucked into this black hole and can somehow get a message through the event horizon, I would be curious to hear your thoughts. It's good to know that this problem is affecting the 'experts' as well, and not just us. I gave those papers a very cursory look, but my field of study is way too far removed from most of those to understand which of those are actually relevant, and I can forget about right.

The social norm thing: yeah, that's probably a risk, particularly on the somewhat multipolar enWiki. I hoped that I could pull a few dozen high-quality sources (leading scholarship, Newspapers of record, recent publications by leading rights organizations) as a sample, but I'm guessing that anything that goes beyond the New York Times won't find the sort of wide-spread consensus required for this, even if it was technologically feasible. FortunateSons (talk) 09:50, 1 April 2025 (UTC)[reply]