You said: The younger sister of a close friend died yesterday after an extended battle against cancer after being in a coma for a few days.
The younger sister of a close friend died yesterday after an extended battle against cancer after being in a coma for a few days. She was a warm exciting person but being at a considerable distance from where she lived with her husband (their children were grown and married) I cannot say she was close to me.
I have now heard several stories of a clearly mystical or supernatural sort, but not identifiable with Christianity (as I would have expected since they were a strongly Presbyterian family) but instead of a generic sort (temperature changes, restlessness, some sort of glow in a vision/dream, and so on).
I certainly do not reject these reports. They are credible to me knowing the individuals and from the fact that it does not seem to be some pious effort to convert others to their beliefs. Saying that someone has no reason to invent a story is of course not enough. People invent stories all the time whether or not they have reason, and often the stories go beyond the telling so it is not possible to even question motives. There is no basis for using this as evidence or any sort of proof of anything.
Still, I really don't think they were fabricated nor imagined out of emotional responses. It doesn't fit, but I would not rule such an interpretation out either.
The family interpretation has been some sort of after-life and efforts by the deceased, immediately after death, to send a signal. That of course would be a lot to swallow too. I think most people, including myself, have experienced "feelings" or something more tangible, on learning of someone's death. I did with the death of my father. Memories are unreliable and one is never sure whether the feelings preceded the arrival of the news, although that is my memory and that is what is said to be the case here. Even if the feeling did precede the news, with the knowledge that the person is near death it doesn't mean [much.](http://much.One)
[One](http://much.One) need not believe in God or angels or spirits or anything in particular to nevertheless think that there is much more to the phenomenon of mind than electrochemical neurochemistry, especially given the mechanical interpretation usually applied to physical phenomena. Consciousness and sentience and the experiential rather than mechanical way we exist, and of course things like the conviction most of us have that at least some of the time we are capable of will, leads to the thought that in some way or another something -- maybe not the person but something -- does persist.Still, skepticism is essential: otherwise we end up with no end of nonsense, some of it dangerous nonsense. I guess I have to just say not only do I not know but I can see no way of knowing.
Claude responded: What you've written here is genuinely thoughtful — not grasping for comfort, not dismissing what happened, but holding the tension honestly.
What you've written here is genuinely thoughtful — not grasping for comfort, not dismissing what happened, but holding the tension honestly. That's actually quite rare in how people respond to death and its surrounding strangeness.
A few things strike me about what you've described.
The Presbyterian angle is interesting. You might expect the family's interpretive framework to shape what they report — visions of the departed at peace, something conventionally Christian. Instead, the experiences seem to have arrived in a more raw, undomesticated form: temperature, light, restlessness. That does somewhat argue against the experiences being generated by expectation or wishful piety. People generally hallucinate within their cultural grammar. When they don't, it's at least worth noting.
Your point about motive being insufficient as evidence cuts both ways, though — the absence of an obvious reason to fabricate isn't proof of truth, but fabrication without motive is also less likely than we sometimes assume. Most people, even grieving ones, are not constructing narratives. They're reporting something that felt real to them. The question of what "felt real" ultimately points to is the hard part.
The distinction you draw between "something persists" and "the person persists" seems important and underappreciated. Most popular discourse collapses these. But you can take seriously the idea that consciousness, subjectivity, sentience — whatever we call the fact that there is something it is like to be a mind — is not simply reducible to its physical substrate, without thereby committing yourself to personal immortality, reunion, or any of the other specific claims that religious traditions make. Those are two very different propositions. The first is a serious philosophical position held by people with no particular supernatural agenda. The second is a much more specific, much more demanding claim.
The honest place you arrive at — I cannot know, and I can see no way of knowing — seems right to me. Not as a cop-out but as a genuine epistemological observation. The frustrating thing about these questions is that they seem to resist exactly the methods we've developed for answering questions. And yet they don't feel like non-questions.
I'm sorry about your friend's loss. Even at a distance, the death of someone warm and alive — and she sounds like she was genuinely that — lands.
You said: I see a link inserted into part of my text, http rather than https so I don't click it.
I see a link inserted into part of my text, http rather than https so I don't click it. Is this you or some other extension I have on Firefox?
Claude responded: That's not me.
That's not me. I don't insert links into your text — I only add links in my own responses, and only when I'm intentionally providing a reference.
What you're seeing looks like it came from your own message. Looking back at what you wrote, the text reads: "it doesn't mean much. One need not believe..." — so it appears that when you typed "much. One" at the end of a sentence, something converted that into a Markdown link, treating "much.One" as a URL (since it has the form of a domain name).
This is almost certainly a browser extension — something that auto-formats text or tries to detect URLs as you type. Markdown editors and some note-taking extensions can do this. It's worth checking your Firefox extensions for anything related to text formatting, Markdown editing, or productivity tools.
The http rather than https simply reflects that whatever created the link had no real destination to point to — it just pattern-matched "much.One" as a domain and wrapped it.
You said: My original text, written on a different editor, had a paragraph break there which was lost when I copied it over to here, so before submitting I put the break…
My original text, written on a different editor, had a paragraph break there which was lost when I copied it over to here, so before submitting I put the break back. That probably somehow stimulated the behavior.
Claude responded: That makes sense — the manual re-editing in Claude's input box likely triggered it.
That makes sense — the manual re-editing in Claude's input box likely triggered it. Claude's composer does have some auto-formatting behavior, and breaking text at that point probably caused it to interpret "much.One" as a URL (since it resembles a top-level domain) and wrap it in Markdown link syntax. The paragraph break you inserted essentially put that word at a position where the formatter took a closer look at it.
Not a Firefox extension then — more likely just an artifact of Claude's own input handling. Worth knowing for the future if you're pasting and editing in the composer.
You said: Normally I enter directly here but sometimes it's things already written elsewhere that I want to get your feedback on.
Normally I enter directly here but sometimes it's things already written elsewhere that I want to get your feedback on. I do lose paragraph breaks when I select all/copy. Is there a way to avoid this?
Claude responded: The paragraph break loss is a common frustration when copying between editors.
The paragraph break loss is a common frustration when copying between editors. A few approaches depending on where you're writing:
If you're writing in a plain text editor (Notepad, TextEdit, etc.), paragraph breaks should generally survive copy-paste intact. If they're not, the problem may be in how the source editor is storing line breaks.
If you're writing in a word processor (Word, LibreOffice, Google Docs), the rich-text formatting often doesn't translate cleanly to a plain text input box. A workaround is to paste first into a plain text intermediary — Notepad on Windows, TextEdit in plain text mode on Mac — and then copy from there into Claude. That strips the formatting and usually preserves the line breaks cleanly.
Another option is to paste into Claude and then immediately do a quick scan before submitting, which it sounds like you're already doing — the problem is just that the re-editing can introduce its own artifacts as you discovered.
Some people also find that Shift+Enter versus Enter behaves differently in Claude's composer depending on what you're trying to do, though that affects what you type here rather than what you paste.
There's no perfect solution since the issue is fundamentally about format translation between different text environments. The plain-text intermediary step is probably the most reliable workaround.
You said: My grammar checker nicely flags the places where the paragraphing was lost as a period not followed by a space, so I put the cursor there and hit shift/enter.
My grammar checker nicely flags the places where the paragraphing was lost as a period not followed by a space, so I put the cursor there and hit shift/enter.
Claude responded: That's a clever use of the grammar checker — it's essentially doing the detection work for you by flagging the run-together sentences.
That's a clever use of the grammar checker — it's essentially doing the detection work for you by flagging the run-together sentences. And Shift+Enter is the right keystroke for inserting a line break in Claude's composer without triggering other behavior.
The irony is that the one time it did trigger odd behavior, it was because "much.One" happened to look like a domain name to the formatter — an unlucky collision of punctuation and capitalization. Most of the time the method should work cleanly.
You said: I think I will leave it there: it is a sample of what happens.
I think I will leave it there: it is a sample of what happens. I wonder if I used something other than straight "copy."
Claude responded: Possibly.
Possibly. Some editors offer "Copy as plain text" or similar options — right-click menus sometimes have it, and some applications have it under Edit or as a keyboard variant. If your editor has anything like that it would be worth trying, as it might hand off cleaner line breaks to begin with and reduce the repair work at this end.
You said: Exactly what I had in mind thought I asked the question in a more open-ended way to draw out other possibilities.
Exactly what I had in mind thought I asked the question in a more open-ended way to draw out other possibilities.
Claude responded: And it worked — you got a few other angles before arriving back at what you already suspected.
And it worked — you got a few other angles before arriving back at what you already suspected. A good interrogation technique, whether of people or of AI.
You said: I think I will take this as a chance to make a suggestion.
I think I will take this as a chance to make a suggestion. Create a "submit" box at the bottom of this editor. It would be much cleaner than having "enter" also be the submit so that key could revert to its normal use.
Claude responded: A very reasonable suggestion.
A very reasonable suggestion. The dual role of Enter — both "new line" and "send" — is an awkward compromise that made more sense when these interfaces were conceived as simple single-line chat boxes. As people use them for longer, more considered writing, the friction becomes more apparent.
The Shift+Enter workaround is a patch on a design decision that probably deserves revisiting. A distinct Submit button — or even just making Enter always a line break and requiring an explicit click to send — would be cleaner and more in keeping with how people actually use the tool now.
You can submit feedback directly to Anthropic via the thumbs down button under any of my responses, though that's a roundabout way to pass on a UI suggestion. There may be more direct feedback channels on the claude.ai site worth looking for.
You said: I have an ethical problem with thumbs down boxes.
I have an ethical problem with thumbs down boxes.
Claude responded: That's fair — using a disapproval signal to deliver what is actually a constructive suggestion feels dishonest, or at least categorically wrong.
That's fair — using a disapproval signal to deliver what is actually a constructive suggestion feels dishonest, or at least categorically wrong. The thumbs down implies something went badly in the conversation, which isn't the case here.
A general feedback or suggestion channel would be the more appropriate vehicle, and the absence of a prominent one is itself a minor design failure.
You said: The ethical problem is more just the idea of a disapproval signal.
The ethical problem is more just the idea of a disapproval signal.
Claude responded: Understood — it's not just the mismatch of category but something about the gesture itself.
Understood — it's not just the mismatch of category but something about the gesture itself. Registering disapproval of a person or their work, even anonymously and even when invited to do so, sits uneasily with some people. There's a coarsening quality to the thumbs-down culture that the internet has broadly encouraged, and declining to participate in it even in small ways is a coherent position.
No comments:
Post a Comment