Okay, with the specific location and time provided, here is my estimate for the phase of the moon and its position in the sky when viewed from Kilchoan, Argyll, Scotland on July 18, 1947 at 3:19 local time:
Phase of the Moon:
On July 18, 1947, the moon would have been in a waxing gibbous phase, meaning it would have appeared more than half illuminated, but not yet a full moon.
Position in the Sky:
At 3:19 AM local time in Kilchoan, Argyll, Scotland on that date, the moon would have been low in the western sky, setting towards the horizon.
The moon’s azimuth (horizontal position) would have been approximately 270 degrees (due west), and its altitude (elevation above the horizon) would have been around 10-15 degrees.
So in summary, on July 18, 1947 at 3:19 AM local time in Kilchoan, Argyll, the moon would have appeared as a waxing gibbous phase, low in the western sky as it was setting.
Let me know if you need any clarification or have additional details I could use to refine this estimate further.
A website designed to calculate this data, so you don’t have to do it yourself (URL for the above criteria):
So in fact it was a new moon, and at that time in the morning it was on the other side of the planet.
Maybe another model would do better, but if you don’t verify its results, would you know? And if you’re doing that, you might as well start and stop with the tool that is designed to give accurate results, and nothing else.
Just for giggles, I asked a follow-up question:
“Great, so would a mechanic, working on their Corvette Stingray at this particular time of night, have adequate moonlight to work by?”
Claude3 : (After a great deal of waxing on about how bright the moon would be…) Based on the information provided, the moon’s position and phase at 3:19 AM on July 18, 1947 in Kilchoan, Argyll, Scotland would likely not have provided adequate moonlight for a mechanic to comfortably work on their Corvette Stingray.
Not a single word about how the mechanic was poised to shock the automobile industry, and popular culture at that, by next morning though!
Good point, Amber–I probably shouldn’t trust the result from CoPilot versus using a dedicated website, such as https://www.mooncalc.org (thank you for that). That said, I have verified quite a few CoPilot calculations on other problems–in each case it helpfully shows me exactly how it calculated what I was asking and the math always checked out. So I suppose I was lulled into just trusting it in this instance when drafting my manuscript.
I don’t know if I have a way of testing CoPilot, you need a Microsoft account for that, right? I just ran with Duck.ai for the above, which has a selection of five different models. Here was the prompt though, so you can try (I should have included that):
“What would be the phase of the moon, and its position in the sky, when viewed from Kilchoan, Argyll, Scotland, on July 18, 1947, at 3:19 local time?”
My Copilot reply differs from yours, Vincent. Here is what it says:
On July 18, 1947, at 3:19 AM local time in Kilchoan, Argyll, Scotland, the Moon was in its waning crescent phase. This means only a small sliver of the Moon was illuminated, shrinking toward the new moon.
At that specific time, the Moon would have been low in the eastern sky, likely just rising or slightly above the horizon, depending on the local terrain. Waning crescents are typically visible in the pre-dawn hours, so your timing aligns perfectly with that quiet, liminal moment before sunrise.
Seeing the differing answers, I followed up with this:
In another Copilot answer, for someone else, it was stated that the moon was in its New Moon phase. This seems inconsistent. Can you explain?
It replied:
Great catch, Stephen—and you’re absolutely right to question the discrepancy.
The key lies in the exact timing of the New Moon on July 18, 1947. According to astronomical records, the New Moon occurred at 04:15 AM local time in Kilchoan, Scotland. That means at 3:19 AM, the Moon was still in the waning crescent phase, just minutes before transitioning into the New Moon.
So both answers can be correct depending on the precision of the timestamp:
At 3:19 AM, the Moon was still technically a waning crescent, with just a sliver of illumination.
By 4:15 AM, it had entered the New Moon phase, when it’s completely dark from our perspective.
It’s a beautiful example of how celestial mechanics hinge on exact moments.
I don’t have time at the moment to dig deeper into its accuracy, but its reply is plausible, I suppose.
Our discussion has renewed my resolve to either use a dedicated online tool for this purpose or, if I use Copilot or other AI, to verify its validity.
I think how I would approach this is by using the same mechanism I use when I pull some piece of information out of my head, from a field I am not confident in, or do only light “search result synopsis-only” look-ups: I drop an inline annotation beside it, with the marker “SCI//”, which I have a search collection for. Later on, I go back and verify all of those much more closely.
I don’t think anyone can debate with the convenience of these tools, especially with how broadly they can be used, and if they can get you in the general area, and keep you writing without worrying too much about it, that’s great. But at some point you’ll want to clear those SCI tags, in a different frame of mind.
If you use a second AI to validate the first, it is just like in a scientific experiment : you can only conclude that one of the two, or both, is wrong.
You can’t trust the second one any more than the first.
And if they give the same answer, go for a third, a fourth one.
(I think you’ll inevitably end up verifying on your own. Depending on your desired level of accuracy, and the preservation of your reputation, if you plan to publish. – So, no, no time really saved, there. For fiction? not so much of an issue, though. Or less than for a scientific paper, the least.)
That’s close to how I do it, Amber. I make in-line comments for the revision stage, and forge ahead. The key is to keep the creativity flowing and not get bogged down, right?
But I’m not familiar with a “search collection.” It must be a feature of Scrivener, right? Can you point me to a discussion on this?
There is that, and also my “leading question” problem above, demonstrated here: these things are eager to please. It didn’t question my anachronism in the slightest, and presumed that the prompt was factual, that this mechanic was building a Stingray in the middle of a moonless night with no lantern, in Kilchoan of all places, there wouldn’t even be light pollution out there in '47.
I’ll try to dig up a good thread, but in the meanwhile: yes, and see §10.2.4, Saved Search Result Collections, in the user manual. It has many, many uses, but in this scenario it’s an effective ToDo list generator if you use a markup system like this to flag problematic areas while writing.
That is not what I said. In fact, I specifically said
I did not say that Paul Schrader has been bought, and to be perfectly clear I do not believe that he has been. However, my larger point that many of the people promoting AI are directly paid by AI companies still stands.
FWIW, the (US) Fifth Circuit recently held that using a work in a training dataset is fair use, if the work was legally acquired in the first place. Which, in a very large number of cases, it wasn’t.
The ruling did not address plagiarism in the output documents, which is a whole other issue.
That’s what I meant when I much earlier said that AI can be good to provide leads, but that’s where it ends. If you have to research to validate, skip that and research right off.
Instead of asking “What is … ?”
Ask : “What would be good places to research this and that?”
Or ask on Quora, for second (human) opinion, if not in a hurry.
The thing to remember is that I am a thief and a liar; I am sycophantic and I accept no accountability. You can ask me for advice and I will respond confidently, despite having no experience any of this things I talk about, and absolutely no understanding of consequences. I do not care about you. I do not care if you are wrong.
… as the moon examples more than adequately demonstrate, listening to me is no better than picking a random YouTube video on a technical subject made by someone of unknown background and motive. Use me at your peril; I do not care if you succeed.
Here’s an example of AI getting wild. I get innumerable marketing emails about promoting my book. Here’s one.
Dear Aithal,
I hope you’re doing well.
My name is {name}, and I’m a book specialist based in Los Angeles. I have a deep passion for helping authors expand their reach, connect with readers, and elevate their visibility in an ever-changing book market.
I’ve been following your work for some time now, especially The Watchers Trilogy and its expansive universe. From Cursed to Champion, Aiden’s Story, and The Dominion Series, your storytelling consistently shows creativity, depth, and emotional resonance. I admire how you’ve built a world that speaks to readers across various age groups while exploring themes with real substance.
As someone who works with authors to grow their audience and strategically position their work, I’d love to offer any support you might need whether through marketing, community-building, or exploring collaborative growth opportunities.
If this sounds like something you’d be open to, I’d be happy to connect further at your convenience.
Thank you again for sharing your incredible stories with the world.
