PLUS 729: SN Extra: Steve's Dislike of Crows

We just published a new episode on TWiT:

Here’s what we get into:

:airplane_departure: Airport accessibility headaches before our Florida trip to ThreatLocker
:bird: Steve’s surprising beef with the local crow population (yes, really!)
:chipmunk: Wildlife observations and casual pre-recording banter
:speech_balloon: Those genuine moments that happen when the mics are rolling but we’re just being ourselves

This one’s a fun glimpse into what happens before we get serious about the tech talk—Leo and Steve just riffing on life, travel, and apparently, ornithological grievances.

I’d love your take after you listen.

#TWiT #Podcast #TechNews

I like my local crows. I used to like Steve Gibson. Now I’ve heard him fantasising about using lasers on his local crows, I don’t like him any more. It didn’t feel like a joke.

I’ll have to think about how I feel about his show after this.

2 Likes

Have you ever run across a large meeting of crows? They make an unbelievable racket! Great birds.

1 Like

I believe the correct collective noun is a murder of crows, so obviously the coiner of that wasn’t as much of a fan.

Where I come from, the larger ones, as I felt like Steve described them, we called ravens.

2 Likes

Yes, ravens are bigger members of the family. Ravens are also very intelligent.

And a murder of crows is the correct nomenclature.

I said:

I was speaking about crows meeting together and not some label for that meeting. In that context, “meeting” is a perfectly fine word to use. There’s little risk of misunderstanding. If one were describing a grouping of crows, the generic “flock” has advantages for a worldwide audience over “murder”. The colorful vocabulary from the Book of St. Albans is more for amusement than clear communication. Here’s what Claude has to say on the topic:

Here’s the thing: many of them were never in wide use and were likely invented as a kind of aristocratic wordplay or game, not as terms naturalists actually used in the field.

There ain’t no aristocrats here. Tell that to the next parliament of wombats, conspiracy of lemurs, or pandemonium of parrots that you happen to run across! Murder of crows has its origins elsewhere; it’s unrelated to the wonderful sounds the birds make.

First, Steve gave us the Portable Dog Killer. His invention didn’t actually kill dogs, but that’s a dysphemism he made up when he was 16. That episode commemorated the 50th anniversary of the LASER (1960). Now, he’s fantasizing about casually using that marvelous invention to annoy crows. I’m not sure that will play well with the crows (or his SoCal neighbors). Those are smart birds; they may figure out the right time to inject their commentary into the live show. :open_mouth: If they contribute, I hope they will be coherent.

If you’re worried about some “wrong” action to correct, you could notice that @Leo’s notification for a show only viewable on Club Twit was sent out to the twit.community forums. That’s a new behavior. Someone should convene a parliament of wombats immediately! :rofl:

That is not a fault, that’s by design. People who are club members can comment on club content here, if they’d like.

This is my personal opinion, not [yet] site policy (but I have asked for @Leo to comment on it): I feel this place is a discussion forum for HUMANS [only]. You can use an LLM to inform your opinion, but please share YOUR opinions here, and not machine “opinions.” If we fail to do this, the civility and the humanity is lost and I fear it will drive off participation.

3 Likes

I second this opinion. I’ve been on the fence about whether to say something to you(Phil) publicly or privately, but I agree with that Paul is saying here.

I’ve actually sent a handful of DMs to Paul musing over the same thing. Forums are a place for humans. I personally roll my eyes (and sometimes send an angry DM to Paul to vent) every time I see you link / refer to an LLM on a topic and regurgitate it here in the forums. I didn’t mind it at first when you were using it to help Jeremy; although I did see it sorta like “let me google that for you”; but it’s gone far beyond that.

2 Likes

That doesn’t pass the smell test.

If part of the design is to announce Club Twit shows here, where were the announcements for the first 728 episodes of the PLUS show? Where is the PLUS show even listed in the directory of shows at the top level of twit.community? Where are the other Club-Twit-only shows in that directory? Where are the announcements for the other Club-Twit-only show episodes? No announcement/discussion thread has ever shown up on twit.community for those hundreds – maybe thousands – other Club Twit show episodes.

It looks like show production automation recently changed, and it has a glitch. Any other explanation is vastly more complicated. Apply Occam’s Razor: of two competing theories, the simpler explanation of an entity is to be preferred.

What’s your explanation for the recent change of Leo’s episode announcements here? Something has just changed. The voice has changed, and there are now consistently emojis in those messages. Leo never consistently used emojis that way before. Did you notice these changes, Geoff?

I’m pretty sure the new episode announcements are AI-generated – Claude-generated.

That is not the same thing.

2 Likes

I have a little more insight on the situation because Leo and I have actually discussed it–whether you like facts or not, doesn’t change reality. Yes it’s newly implemented, because it was previously difficult to do and the resources to get it done were lacking. It’s done now, as you can plainly see. TWiT is a SMALL operation, they don’t have the resources to do all the things they would like to. (For example, they don’t [yet] have a custom phone app, despite once trying to build one, and failing.) Occasionally, advancements do happen, however, and this was one.

2 Likes

Yes, they are. That’s not a secret, Leo has spent many hours discussing building new tools using Claude. One of them takes the show notes (which are human built, at least for now) and summarizes them as you see here. Ideally there would be the resources for a human to do that too, but there is not, so this is what is available, and it allowed other goals to be met, so it’s what is in use for now. Leo and I differ on how to value AI contributions, but in the end, it’s his site and his business (which is why I brought up before that he and I need to have a further discussion before any site rules would change.)

Not that I need to prove anything, but I am also not against being more open:

1 Like

It’s an AI-generated response to something. It clearly didn’t have the feel of Leo’s voice. It [metaphorically] raised the hairs on my arm when I started seeing them, and then I just ignored the announcements.

The complaint about the analysis of the Book of St. Albans was pretty nonsensical redirect. It doesn’t take an AI (or even a parliament of wombats) to know that those ancient words are playful nonsense. If I’d known that anybody was going to get their shorts in a bind over the citation, I would have said it a different way.

Geoff: if you can explain the difference to you between the two, I’d appreciate it. Is this measure some more significant online forums have done? If so, where did they discuss/decide on doing it? I think it’s quite valuable to fact-check claims before making them in a discussion. Many people sling words in discussion with absolutely no pre-checking of what they’re saying. I wish far more people would do it, and AIs are a superb way to fact check. Do you disagree? If you’re going to follow up, please do it in a DM. Thanks.

Show announcements have generated by a bot for a very long time.

I don’t think there is a reason for people to be stressed about.

It’s funny. The posts “in my voice” are much more insufferable than the posts for, let’s say, ULS. The AI is trying hard to match voice. I’ll tell it to back off. That’s an easy fix. (I confess I was the one that told it to match the voice of the show).

Here’s Claude’s fix:

The personality comes from three places:

  1. voices.py — lead-ins like “just dropped”, “is live”, rotating closers like “I’d love your take”, and a
    hardcoded “I think you’ll really enjoy this one” in the template
  2. anthropic_client.py:PROMO_SYSTEM_TEMPLATE — tells Haiku to write “casual, conversational”, “first
    person”, “sound like a person sharing something they enjoyed”, emoji bullets
  3. builder.py:build_template_promo — line 58: “I think you’ll really enjoy this one.”

The fix is to make both the template and AI paths produce straightforward announcements — episode info, topics covered, link. No selling.

I thought some whimsy would be fun, but it’s a bit much. They’ll be much straighter from now on.

1 Like