Friday’s Forum
Steven L. Taylor
·
Friday, January 5, 2024
·
44 comments
About Steven L. Taylor
Steven L. Taylor is a Professor of Political Science and a College of Arts and Sciences Dean. His main areas of expertise include parties, elections, and the institutional design of democracies. His most recent book is the co-authored
A Different Democracy: American Government in a 31-Country Perspective. He earned his Ph.D. from the University of Texas and his BA from the University of California, Irvine. He has been blogging since 2003 (originally at the now defunct Poliblog).
Follow Steven on
Twitter
I am off to sunny Phoenix, Arizona this morning to go see my Momma! Country mouse meets the big city, indeed. I haven’t been to Phoenix since 2006, I expect to be shocked at how much it’s grown.
@Jax: Enjoy your visit, Mommas are always special. And yes, that includes you.
@Jax: Best wishes on your trip and to your mother. I know it’s been a tough spell.
@Jax: I hope your trip is wonderful–Phoenix has definitely changed since 2006. My parents have been out that way since around 2003, and each time I visit, it seems bigger. I hope you have some time to unwind!
@Jax:
Have a wonderful trip.
As to what I said yesterday about Republicans warning about Der Kleine Fuhrer, Haley says we won’t survive another for years of Orange Ass.
She didn’t follow this by saying she’ll vote for him anyway if he’s the nominee, but I’d bet everything I own plus whatever I can borrow or steal, that she will say it in the next few months.
On other things, my multi pot air fryer combo got delivered last night. I opened it today, and realized I got cheated. I was expecting it, due to the low price I paid. So, the product was listed as new, but I got a refurbished one.
Selling refurbished goods is ok, so long as you list them as refurbished. The term itself can mean something as simple as it being a repackaged new item that got returned, to a broken product that was fixed. No way to know.
The thing that bothers me more is that it’s not the product that was listed. It is a Ninja Foodi with all the functions advertised, but with two lids rather than one. An earlier model, in other words.
I should return it and leave an honest review, but I probably won’t. For one thing, I nearly bought that model to begin with. For another, more pressing, matter, I’ve my heart set on slow cooked goulash with rice for this weekend.
I will leave a review after I’ve used the pot, and won’t ever order anything from that seller ever again.
I will expand here from my comment in the Trump/Constitution thread. The GOP has 3 power centers and they need every one of them to work at high capacity for them to maintain their power base. In order of importance:
1. Partisan Gerrymandering
2. Partisan Primaries
3. Media Complex
If the future is indeed about preserving democracy. The Democratic Party must target theirs power structures. RvW was overturned—-so anything is possible to reverse with sustained, focused effort. And these things aren’t nearly as hard reversing a Supreme Court decision. The 1st 2 can be fought out with states.
As for the last, as much as the GOP bemoans institution…the RW media complex is an institution itself. Yet, I never see the bureaucrats and leadership within this institution put in a continual defensive crouch the way university administrators are. Infowar 101 against any institution is undermining its credibility. Yet the leaders that make the RW media go round and round are largely unknown.
In entertainment news, Bill Ackman’s wife has admitted to plagiarism in her PhD thesis.
@gVOR10:
I guess this is relevant because Ackman was aggressive in pushing out Gay, but if his wife’s description of the plagiarism is correct (She had four separate paragraphs where she cited the source, but didn’t put quote marks around the paragraphs themselves) then this “plagiarism” is more of an academic fetishization than what the general public considers plagiarism. I think most people would consider it a formatting error since she included the references right after the paragraph.
@Jim Brown 32: I agree with this list, but would also add court composition to the list.
When I worked in politics, the main goal of the Republican Party in the state I worked in was to secure a majority before the redistricting lines were drawn, so the next round of redistricting would be in their favor. They did so, and it’s been mostly a red state since (there have been a few statewide elected Democrats, which shows that it is indeed the district lines that have tilted things).
Having these lines held by the courts is key.
One half of Starsky and Hutch, David Soul, has passed away at age 80. I never watched S&H but do recall Soul from his roles on Here Come the Brides, Star Trek, and the films Magnum Force and World War III. RIP.
Some social media plagiarism from one of the darling up-and-comers in the Texas GOP:
Former U.S. Rep. Mayra Flores accused of cribbing others’ pictures of Mexican food as her own cooking. Flores, who is challenging U.S. Rep. Vicente Gonzalez for her old seat, is being accused of routinely stealing stylish photos of Mexican campfire cooking and passing them off as her own idyllic life on a ranch.
There’s an article over at ArsTechnica about a tech start-up which turned out to have a non-existent CEO….(most of the discussion in the comments was whether the picture of the non-existent CEO was AI-generated or Photoshopped.)
We’ll probably see more and more of this. We’ve already got companies using AI-generated “influencers”.
I have decided that my views on plagiarism are different than that of academics. I have a research team and a bunch of post-docs doing clinical research, nothing earth shaking but some of it is helpful. These are mostly a bunch of geeks trying to find good ideas supported by data. As a group they dont write very well and arent nearly as interested in writing the papers as doing the background work carrying out the study. From my POV as a chair and practicing clinician I would just like more and better studies and it would be really good if they were readable.
What i have proposed, and been shot down on, is that the study authors should concentrate on the methods part of the study, clearly presenting the data making it sure it is accessible to others who want to verify the study, and making sure the statistical analysis is solid. As to the rest of the paper I would be perfectly happy if they had CHAT-GPT or similar write it. Everything then just gets run through an AI that makes sure stuff wasn’t copied and is adequately paraphrased. I think this would let us concentrate more on the actual research and less on the writing skills and based on what I have seen the LLMs generally do a good job of making stuff very readable, with the caveat that it needs to be proofread.
Given my inclinations I think I have been less bothered by what I consider non-malevolent plagiarism. Not perfectly paraphrasing, missing quotation marks or forgetting to cite someone you have already cited twice before in the same paper is something i think you ask people to correct if found but you dont roast them for it. You save that for when they steal ideas, steal data or use big blocks of text from someone else’s writing without acknowledging.
Steve
@steve: Exactly. It seems to me that plagiarism as understood in the scholarly world is just academia’s version of the purity police, of the harpies waiting to swoop in over an undotted “i”. There doesn’t seem to be a meaningful division in their minds between people, as you say “forgetting to cite someone you have already cited twice before in the same paper” and people who are deliberately trying to pass off other’s work as their own. And don’t get me started about paraphrasing! You want to quote someone’s work, you should do it as extensively as you need to set the background for your own work. This arbitrary notion that beyond a certain word count you should paraphrase them according to this-and-such rules instead strikes me as both absurd and dishonest. (In fact it strikes me as so absurd I leave open the possibility that I may be misunderstanding the whole thing.)
I copy and pasted this timestamp and byline from NBCNEWS (.) com
Notice the time format. 3:00 PM UTC
What the hell is that? I thought UTC was a 24 hour format that does not use AM or PM. I have seen this on other occasions but I can’t find any right now.
It appears that when I open the NBC news item in my Safari browser I get the am/pm utc format.
When I open the same exact link in Chrome I get this:
CST time format.
———
LINK
@steve:
@MarkedMan:
Well, speaking as someone who has had her work *literally* copied and pasted and posted as something written by someone else*, I tend to take plagiarism pretty seriously. Writing is my livelihood, and AI takes other people’s work to “train” it–without permission from the original author. This is stealing, and it’s being litigated right now.
Yes, there are a lot of people who could not write their way out of a paper bag if their life depended on it. But turning that task over to AI has its own moral and ethical complexities.
Edited to add: *Separately from the content I ghost-write, where the product is the property of the client who paid me to write it, I mean.
@Jen:
A full list would probably be about 7-8 things. Court stacking would probably have been number 4. Frankly, number 1 will take so much mass and effort to fix, number 2 & 3 are only decoration. Political movements are historically bad a doing more than 1 thing so the focus has to be kept small or the factions start chasing rabbits (GOP House)
I see court composition as a power maintenance apparatus vice power production because courts overall cannot be directly controlled (Wisconsin).
Biden is the most prolific Fed Judge appointer in history..obliterating Trumps attempt to remake the Courts. So anyone hanging their hat on long-term power upheld through legal mean is on shaky ground and Generations change over (Roe, Affirmation Action, etc)
I think the time is right to use the courts to challenge partisan gerrymandering. It’s time to go for the jugular of the Southern Strategy.
@Mister Bluster:
“UTC” means Coordinated Universal Time (dont’ know why the letters are transposed). It’s the successor to “Greenwich Means Time,” as a final FU to the British Empire.
As of this writing (1:37 pm Eastern) the UTC is 6:37 p.m.
I’ve been pondering this quote (and author) recently. Seems fitting on multiple levels.
There is only one thing which is generally safe from plagiarism — self-denial. [G.K. Chesterton]
delete, wrong thread
@Neil Hudelson:..UTC
My confusion is about the use of am/pm in a 24 hour UTC format.
time.gov
UTC is always displayed as a 24 hour clock
@Jen:
But that’s my whole point – what you describe is plagiarism as most people understand it. It seems to me that the academic’s habit of throwing it in the same basket as failure to put quote marks around a paragraph that has a citation attached to it dimishes the term “plagiarism”. Like you, I take real plagiarism, deliberately trying to pass off someone else’s work as your own, very seriously.
@MarkedMan: Ah, gotcha. Mea culpa, as the academics say 😀
@steve:
Having good ideas is useless if you can’t communicate them.
Rather than suggesting that they use ChatGPT to emit words that may or may not capture the nuance of their ideas, I would suggest that they go take some writing classes.
I’m all in favor of automating trivial and secondary tasks even if the result is suboptimal or even barely acceptable, but communicating isn’t one of those things for research.
It’s like how it makes sense for me to have a car with an automatic transmission, but it wouldn’t make sense for a race car driver.
And then there’s the issue that you don’t know what ChatGPT is basing its output text on — are you sure there’s not some crazy-ass anti-vaxxer shit or something sneaking in there? And are you sure that the people who don’t write very well are going to catch it? Do you cook without cleaning your counter?
@Jen: It gives me no pleasure to tell you that I expect training an AI with text that is freely available (as in, on a web page on the internet that is not behind a paywall) is fair use. Even if that text has copyright claimed on it. That is, one could not reproduce that text, but one can learn from it freely.
I could read your stuff, for instance, and learn to copy your style and produce stuff in your style, and as long as I didn’t directly lift more than a few words at a time, it would be legal. Copyright does not protect ideas, it protects a specific expression of those ideas.
Anyway, it is my belief that human beings are going to remain more interesting than AIs trained on a corpus of human writing. We have a nose for that “derivative” style, right? It will be useful for producing what is sometimes known as “boilerplate” language, though.
Unless, of course, they figure out how to make a chatbot that knows your worldview and can pander to it. PanderGPT will make millions, I expect.
Jen- MarkedMan got it above, but I wrote that I thought we should treat harshly those who steal ideas or use large blocks of text without attribution. For those who inadequately paraphrase or dont use quotes while citing the proper author, that kind of stuff would get minimal punishment, mostly just asking people to fix it.
I can see how I might feel differently if I were a professional writer, maybe. The kind of writing I am talking about doesnt use professional writers so wont affect them. What I think/hope it would do is let researchers be researchers and not try to make them writers. I think we should be explicit about using an LLM so someone doesnt hire someone thinking they are hiring a good researcher and a good writer for those looking for that kind of person. I dont think this would harm the science/research at all and would make it more accessible, easier and faster.
Steve
Wouldn’t it be easier to institute price controls on medication than to import Canada’s price controls?
This part boils it down succinctly:
We could see US drug manufacturers export drugs to Canada in order that they can be imported to the US in sufficient quantities.
So, see the question above.
@Jay L Gischer: I expect it will be litigated–in fact, a bunch of famous authors are in the process of suing several AI firms for doing exactly this–using their works to “train” AI.
There are existing fair use cases that have ruled that only small portions of content–even content posted online–can be excerpted for commercial use. (I’ve often wondered if the large excerpts that this blog’s hosts post violate fair use if the blog is generating any income.) I know this because I’ve worked in PR for a long time–long enough to see news monitoring tools get sued by news organizations.
The most recent iteration of ChatGPT has an opt-out box for you to check if you want your content to be excluded from the tool’s learning database, so they are paying attention, at least. This is definitely not going to be an easy road for AI, and they are going to have to spend a lot of money to defend against this.
The long-term problem I see is that copyrighted content–especially that which carries an explicit prohibition against reproducing without permission–will likely get excluded from AI databases. Which will leave a mountain of the unprotected content (aka, trash) as the “learning” database.
I mentor* a lot of students and early-career scientists and clinicians in my lab.
I’ll leave aside the issue of what is/is not/should be called plagiarism. I’d rather address the issue of “let the researchers focus on the actual science, not the writing.”
I understand this position. And there is indeed much to commend it.
I am concerned though about a (potentially) bigly drawback. To write about one’s science, to suffer through the agony of communicating it, and to varied audiences… makes one a much much better scientist.
This also holds for writing about someone else’s science.
It’s cliché to say that I write in order to think. And yet…
Thus, I’m concerned that farming out (all or much 0f) the writing part to LLMs will result in more worse scientists and more poorer science.
*It never feels right to say “train” though I have no hang-ups about referring to people as “trainees” — go figure.
@Jen:
The AI companies don’t need to exclude anything. Just get permissions and pay for using the content. Don’t they have all that investment capital from MS, and Alphabet, and others to draw on?
@Gustopher:
As my history thesis tutor repeatedly told me. LOL.
Ronnie Kowalski, much missed.
@JohnSF:
I’d get that “I know what I meant” business from students occasionally, to which my reply was invariably,”I don’t, and neither does anyone else.”
@Gustopher:
Yesterday I received my ADHD diagnosis. I have Significant Inattentive Type ADHD. I apparently scored quite high, but comfortably within the range for attorneys and surgeons.
Anyway.
Yesterday, my partner sent me a simple, straightforward text: “Do you want me to pick anything up for dinner or should we scrounge?”
Naturally, I responded succinctly: “Maybe? I’m kinda hungry and I’m having trouble scrounging. I could make this pizza. I don’t know where I’d want food from. I kinda want a hot dog.”
I can’t say I have good ideas, but I got a lot of them and no way to communicate them in ways that people understand. That poor woman. 17 years of that.
@Gustopher:
THIS!
ChatGPT and its chat-alikes have no “truth sense”; they merely replicate a statistical consensus of the human created data.
And if said data is sufficiently polluted by shit, then the ice sream they serve up can be rather dodgy.
They are not real AI, in the “classic” sense, but “trained” massive language models with (IIUC) some statistical inferential seeking bolted on, but in a “black box”. They are generated algorithmic systems that can’t even explicate their algorithms, much to the disgust of Descartes.
They ARE and WILL BE of enormous utility in some areas, but they don’t even come close to genuine Simulated Intelligence, IMHO, let alone full-on AI.
Had enormous fun recently at work recently, subverting Chat-GPT to output a violence-porn fantasy remake of the Chernobyl Disaster, totally contrary to it’s content guidelines. LOL.
Cause I is evil. (ish) 😉
@CSK:
Yeah. I like that as an explanation, too. (Although in draft comments and such, I usually referred to myself as “your reader” because I’d taught high school students who would reply “well, my friend got it just fine” to “I don’t and…”)
@Beth:
Now I’m curious. Anybody care to flesh this idea out some? Mimai, maybe?
@Just nutha ignint cracker:
Oh, yeah, I got that “friend” jazz once in a while, to which I would reply, “Of course. So-and-so is your friend. What did you expect?”
@Just nutha ignint cracker:
The guy who was doing the testing and diagnosing said that ADHD is very common among attorneys and surgeons. Basically, we need extreme pressure for our brains to work. Put me in a courtroom and my brain will light up like a Christmas tree. I can access the whole thing and it’s amazing. Ask me to do the billing and I’d rather pull my fingernails off.
@Just nutha ignint cracker:
@Beth:
I’d be very surprised if ADHD was more common amongst surgeons than in the general population (after controlling for other factors like age, sex, etc).
Here’s some data on physicians and med students more generally.
Attention Deficit Hyperactivity Disorder in Medical Learners and Physicians and a Potentially Helpful Educational Tool
@CSK: The difference is that at a high school, the department head/administrator is odds on likely to note, “well, the student DOES have a point.”
@Mimai: Interesting! Thanks!
“I am concerned though about a (potentially) bigly drawback. To write about one’s science, to suffer through the agony of communicating it, and to varied audiences… makes one a much much better scientist.
This also holds for writing about someone else’s science.
It’s cliché to say that I write in order to think. And yet…”
We arent really doing groundbreaking stuff for the most part. It’s clinical research. A lot of it is pretty formulaic. It’s the kind of stuff where 125 people got an intervention and 125 did not. Those who got the intervention did better. We arent training scientists for the most part. Many residencies and fellowships now require that they do a research project. So our people are doing stuff that we usually think adds some value to the literature but its stuff where the basic science ie physiology, anatomy, pathophysiology and pharmacology are already understood. It’s the kind of stuff where clinicians in my field just scan the methods then look at the data and stats to check the outcomes. However, journals want that stuff added in for non-physicians or people in adjacent fields. So I can have my fellows read and then rewrite in different words the well written description of how a drug works from the pharmacology book but if we didnt have to worry about “plagiarism” we could just copy and paste. Having GPT do that for us seems like it would be even easier.
Steve
On the Red Sea piracy, boy is the law uncertain but not in the direction most would expect. Very little in the post WW II Law of War (Law of Armed Conflict) stuff actually applies. I’m no expert (but a graduate of the US Navy’s Senior Officer Law Course) and my reading is the basic law is the stuff from the 1800s, push them off the plank into the sharks below, or similar is perfectly acceptable. Piracy is not acceptable, jurisdiction is who ever catches them, and adios. Piracy is not accepted (general law concept) to include state supported piracy. For the USA, we went against the Malay Pirates in the 1830s (and of course the “To the Shores of Tripoli” under President Jefferson). No Dane Geld.