caption updates

1 files changed, 93 insertions, 65 deletions

diff --git a/2025/captions/emacsconf-2025-llm--emacs-editors-and-llm-driven-workflows--andrew-hyatt--answers.vtt b/2025/captions/emacsconf-2025-llm--emacs-editors-and-llm-driven-workflows--andrew-hyatt--answers.vtt
index f35c2d27..849f7619 100644
--- a/2025/captions/emacsconf-2025-llm--emacs-editors-and-llm-driven-workflows--andrew-hyatt--answers.vtt
+++ b/2025/captions/emacsconf-2025-llm--emacs-editors-and-llm-driven-workflows--andrew-hyatt--answers.vtt
@@ -1,5 +1,7 @@
 WEBVTT
 
+NOTE Q: My biggest question with AI code editors trying to integrate with Emacs is -- are the AI code editors able to read unsaved buffers and not just saved files?
+
 00:00:00.000 --> 00:00:01.759
 So let's, I'm just going to answer
 
@@ -165,6 +167,8 @@ But yes, that's what I think about.
 00:02:16.600 --> 00:02:20.319
 that interesting questions about unsaved buffers.
 
+NOTE Q: Personally I don't agree with the comment you made about VS Code usage dying out because I see companies/products pushing for tightly-integrated VS-Code agents/products like Windsurf. Thoughts?
+
 00:02:20.320 --> 00:02:22.239
 The next question is,
 
@@ -172,7 +176,7 @@ The next question is,
 I don't agree with the comment you made
 
 00:02:23.800 --> 00:02:25.199
-about VS code usage dying out
+about VS Code usage dying out
 
 00:02:25.200 --> 00:02:26.719
 because I see companies and products
@@ -264,6 +268,8 @@ I think it's gonna be the opposite, but I guess we'll see.
 00:03:43.200 --> 00:03:47.759
 Like, let's reconvene in a year and see what happens.
 
+NOTE Q: Do you have any thoughts about the environmental cost of using LLMs - either the training of models we can download and use locally, or the larger, commercial models used from the cloud?
+
 00:03:47.760 --> 00:03:49.159
 Uh, the 3rd question answer,
 
@@ -408,6 +414,8 @@ So I think it's wise to be cautious,
 00:06:06.120 --> 00:06:09.079
 but I think it's okay, I think, at least for personal use.
 
+NOTE Q: I must say, I liked your conclusion, but I differ insofar as you said that VS Code differ from Emacs because the former is not as easy to adapt as the latter. Why should Microsoft not adapt VS Code as we adapt Emacs for the new era of coding? And why would VS Code be harder hit? Could you please elaborate on this point?
+
 00:06:09.080 --> 00:06:13.159
 The next question is another,
 
@@ -538,10 +546,10 @@ which people are really just using to edit code
 and if people find it less useful to edit code.
 
 00:08:16.800 --> 00:08:20.919
-I think it's VS code will be harder hit than emails
+I think it's VS Code will be harder hit than emails
 
 00:08:20.920 --> 00:08:24.239
-because that's its whole like that's in the name
+because that's its whole, like, that's in the name,
 
 00:08:24.240 --> 00:08:25.559
 like the whole reason for it
@@ -556,7 +564,7 @@ So I think that it's it's vulnerable
 in a way that Emacs isn't
 
 00:08:31.960 --> 00:08:34.519
-just because emacs is so very...
+just because Emacs is so very...
 
 00:08:34.520 --> 00:08:40.119
 you know, it's, it could do so many things
@@ -595,11 +603,13 @@ Is that touching on the point?
 Yeah, that's a great way to say it.
 
 00:09:09.080 --> 00:09:12.439
-Thank you. Thank you, Colin. Yeah.
+Thank you. Thank you, Corwin. Yeah.
 
 00:09:12.440 --> 00:09:14.039
 Thank you. Thank you for that question.
 
+NOTE Q: Do you think that we are falling behind in productivity as Emacs users? Compared to all these VSCode forks that have 1000 buttons and textboxes everywhere (i.e. much richer UIs which are basically webpages).
+
 00:09:14.040 --> 00:09:18.759
 Do you think we're falling behind in productivity as Emacs users
 
@@ -774,6 +784,8 @@ I shouldn't be doing so much active listening.
 00:12:12.280 --> 00:12:17.479
 No, no, I appreciate your input.
 
+NOTE Q: I've been using Claude Code extensively. I recently switched to Agent Shell with Claude Code. Have you tried it, what are your thoughts?
+
 00:12:17.480 --> 00:12:23.079
 OK, next is I've been using Claude Code extensively.
 
@@ -799,10 +811,10 @@ If Agent Shell did exist,
 I probably would have demoed it as well.
 
 00:12:41.240 --> 00:12:45.359
-Agent Hell is great in the sense of it's
+Agent shell is great in the sense of it's...
 
 00:12:45.360 --> 00:12:53.719
-It does use comment, which is the way that I think all Emacs users
+It does use comint, which is the way that I think all Emacs users
 
 00:12:53.720 --> 00:12:57.039
 would prefer to interact with something like Claude Code,
@@ -853,13 +865,13 @@ It's progressing rapidly.
 But it's not as rich in functionality
 
 00:13:37.480 --> 00:13:40.119
-as using quad code directly.
+as using Claude Code directly.
 
 00:13:40.120 --> 00:13:44.839
 On the other hand, because it's letting Emacs be Emacs
 
 00:13:44.840 --> 00:13:49.239
-and using comment, it's a much better experience
+and using comint, it's a much better experience
 
 00:13:49.240 --> 00:13:50.919
 to actually give instructions.
@@ -901,7 +913,7 @@ in order to me, I think there's
 kind of the sweet spot that I like.
 
 00:14:23.520 --> 00:14:26.639
-Um, but HHL is a great step forward
+Um, but agent-shell is a great step forward
 
 00:14:26.640 --> 00:14:29.879
 and I think it's, uh, it's quite good to use.
@@ -909,6 +921,8 @@ and I think it's, uh, it's quite good to use.
 00:14:29.880 --> 00:14:32.119
 And I, I personally use it a lot.
 
+NOTE Q: In terms of agent selection, what has your experience been with different agents, and have you had any success with hosting your own models and using open weights?
+
 00:14:32.120 --> 00:14:40.479
 Um, OK, so in terms of, next question,
 
@@ -937,7 +951,7 @@ I think Claude Code is, most people I know
 would say Claude Code is probably,
 
 00:15:03.480 --> 00:15:07.479
-sorry, Cloud is probably the best for coding right now.
+sorry, Claude is probably the best for coding right now.
 
 00:15:07.480 --> 00:15:09.919
 Gemini can be very hit and miss even with 3.0,
@@ -1027,7 +1041,7 @@ just because they just have more horsepower,
 they can churn through those tokens a little quicker.
 
 00:16:38.680 --> 00:16:44.719
-So, I'll just break in here to say,
+[Corwin]: So, I'll just break in here to say,
 
 00:16:44.720 --> 00:16:46.239
 we've got about 7 minutes left
@@ -1050,11 +1064,11 @@ to jump in and prep with the next speaker,
 00:16:58.680 --> 00:17:00.359
 but you'll be able to keep going
 
-00:17:00.360 --> 00:17:02.599
+00:17:00.360 --> 00:17:04.119
 as long as you have the steam for it.
 
-00:17:02.600 --> 00:17:06.159
-Yeah, I think we have 3 questions. Yeah, thanks.
+00:17:04.120 --> 00:17:06.159
+[Andrew]: Yeah, I think we have 3 questions. Yeah, thanks.
 
 00:17:06.160 --> 00:17:08.239
 I think we have 3 questions.
@@ -1062,8 +1076,11 @@ I think we have 3 questions.
 00:17:08.240 --> 00:17:09.719
 Let's see if we can get through them
 
-00:17:09.720 --> 00:17:17.359
-all in that time period. OK, this one is interesting talk.
+00:17:09.720 --> 00:17:13.267
+all in that time period.
+
+00:17:13.268 --> 00:17:17.359
+OK, this one is interesting talk.
 
 00:17:17.360 --> 00:17:20.439
 I'll start by asking it for everything, but is it editing?
@@ -1077,6 +1094,8 @@ So yes, let us all ask, but is it editing?
 00:17:29.720 --> 00:17:33.439
 All right. I can move on to the comment area.
 
+NOTE Q: I'm reading angst in your thinking about AI/editing.  What are you excited about?
+
 00:17:33.440 --> 00:17:40.319
 I'm reading angst in your thinking about AI editing.
 
@@ -1108,19 +1127,19 @@ And I think there's lots of opportunities,
 clever opportunities to do things
 
 00:18:07.840 --> 00:18:11.079
-we couldn't have thought of Things that are useful,
+we couldn't have thought of... Things that are useful,
 
 00:18:11.080 --> 00:18:14.919
 but in ways that are not super obvious to us,
 
-00:18:14.920 --> 00:18:18.519
+00:18:14.920 --> 00:18:17.019
 and I think I'm still excited
 
-00:18:18.520 --> 00:18:23.839
+00:18:17.020 --> 00:18:22.939
 about the possibilities of using them in ways that are super helpful
 
-00:18:23.840 --> 00:18:29.319
-and different than. normal. I'll give you an example.
+00:18:22.940 --> 00:18:29.319
+and different than normal. I'll give you an example.
 
 00:18:29.320 --> 00:18:33.199
 This is something that I intend to, I think,
@@ -1159,7 +1178,7 @@ Yes. There are some smart people who remember this
 but I am not one of them.
 
 00:19:11.200 --> 00:19:13.639
-And so I think like something like this is like you just type out,
+And so I think, like, something like this is like, you just type out,
 
 00:19:13.640 --> 00:19:17.279
 find me this file, and it will substitute
@@ -1174,7 +1193,7 @@ I think this is, there's a lot of little,
 little tweaks you could do like, you know, if you want the AI,
 
 00:19:24.320 --> 00:19:26.679
-it could be there for you. And it will help you.
+it could be there for you, and it will help you.
 
 00:19:26.680 --> 00:19:27.559
 And if you don't want it,
@@ -1183,7 +1202,7 @@ And if you don't want it,
 it's not going to get in your way.
 
 00:19:28.920 --> 00:19:30.639
-And I think this is where eMath can really shine.
+And I think this is where Emacs can really shine.
 
 00:19:30.640 --> 00:19:33.079
 It can really take advantage of LLMs,
@@ -1200,6 +1219,8 @@ So thank you for that great question.
 00:19:43.920 --> 00:19:47.919
 And then the final question. Yep.
 
+NOTE Q: Why does it matter to have a richer UI? All that is left is basically writing and getting the results.
+
 00:19:47.920 --> 00:19:50.839
 This final question is, why does it matter to have a richer UI?
 
@@ -1216,7 +1237,7 @@ about Emacs not having a richer UI before,
 but I think it does matter a lot for all sorts of things.
 
 00:20:03.160 --> 00:20:07.159
-It's hard to kind of explain succinctly
+It's hard to kind of explain succinctly,
 
 00:20:07.160 --> 00:20:08.519
 because I'm talking about UI
@@ -1225,28 +1246,28 @@ because I'm talking about UI
 and I'd have to show you things.
 
 00:20:09.760 --> 00:20:14.599
-But it should be just something like oh I have an error
+But it should be just something like, oh I have an error,
 
 00:20:14.600 --> 00:20:15.799
 and I'm using flymake and I'm,
 
 00:20:15.800 --> 00:20:17.759
-I'm using the you know I have options
+I'm using the... I have options
 
 00:20:17.760 --> 00:20:19.839
 where it'll show me the error in line
 
 00:20:19.840 --> 00:20:22.639
-by like underlining things and having a little message,
+by underlining things and having a little message,
 
 00:20:22.640 --> 00:20:24.079
-but like, you know what that message
+but like, you know what, that message
 
 00:20:24.080 --> 00:20:26.399
-doesn't appear quite right, a lot of the times,
+doesn't appear quite right a lot of the times.
 
 00:20:26.400 --> 00:20:30.159
-or here's another one like. I program in Python a lot.
+Or here's another one like. I program in Python a lot.
 
 00:20:30.160 --> 00:20:33.479
 And Python, it's super hard to program in
@@ -1275,11 +1296,14 @@ And so you kind of have to hack it in.
 00:20:47.760 --> 00:20:49.479
 And there's lots of ways to mess it up.
 
-00:20:49.480 --> 00:20:52.799
-And when editing, you'll find yourself messing this thing up.
+00:20:49.480 --> 00:20:52.156
+And when editing, you'll find yourself
 
-00:20:52.800 --> 00:20:57.159
-regularly. So it doesn't look quite clean.
+00:20:52.157 --> 00:20:55.260
+messing this thing up regularly.
+
+00:20:55.261 --> 00:20:57.159
+So it doesn't look quite clean.
 
 00:20:57.160 --> 00:20:59.079
 And like, there's little artifacts,
@@ -1290,17 +1314,17 @@ or, you know, there's little ways that it,
 00:21:01.040 --> 00:21:02.359
 it kind of gets things wrong,
 
-00:21:02.360 --> 00:21:06.279
-or you can, you can get things wrong with it.
+00:21:02.360 --> 00:21:03.939
+or you can get things wrong with it.
 
-00:21:06.280 --> 00:21:10.039
+00:21:03.940 --> 00:21:07.519
 So I think that, like,
 
-00:21:10.040 --> 00:21:11.879
+00:21:07.520 --> 00:21:13.619
 there's a lot of issues with that sort of thing.
 
-00:21:11.880 --> 00:21:15.759
-And, and also, like, you know,
+00:21:13.620 --> 00:21:15.759
+And also, like, you know,
 
 00:21:15.760 --> 00:21:17.719
 what if you want to do something like play a video inline,
@@ -1308,23 +1332,23 @@ what if you want to do something like play a video inline,
 00:21:17.720 --> 00:21:19.559
 like, I don't know, you might should be able to do that,
 
-00:21:19.560 --> 00:21:20.839
+00:21:19.560 --> 00:21:21.599
 you might should be able to do anything.
 
-00:21:20.840 --> 00:21:23.799
-But right now it just can't and I think
+00:21:21.600 --> 00:21:23.799
+But right now, it just can't. I think
 
 00:21:23.800 --> 00:21:24.999
-a lot of the reason as well
+a lot of the reason as well...
 
 00:21:25.000 --> 00:21:26.519
-we you know we wanted to be compatible
+you know, we wanted to be compatible
 
 00:21:26.520 --> 00:21:29.919
-with you know TRS 80 machines or something like that
+with TRS 80 machines or something like that.
 
 00:21:29.920 --> 00:21:33.159
-and it's this is important this really is important,
+This is important, this really is important,
 
 00:21:33.160 --> 00:21:34.719
 but I hope there's some way
@@ -1333,7 +1357,7 @@ but I hope there's some way
 that we can kind of eventually figure out
 
 00:21:36.080 --> 00:21:39.279
-how to get the best of you know both compatibility and.
+how to get the best of both compatibility and
 
 00:21:39.280 --> 00:21:44.879
 more modern UIs. So, you know, we can have more modern UIs
@@ -1389,6 +1413,8 @@ that are working behind the scenes.
 00:22:23.320 --> 00:22:25.079
 So thank you all for putting this together.
 
+NOTE Wrapping up
+
 00:22:25.080 --> 00:22:29.199
 I'm so happy that we all are here. We care about Emacs.
 
@@ -1405,22 +1431,22 @@ this really remarkable achievement.
 Like it's amazing that it exists. It continues to exist.
 
 00:22:41.800 --> 00:22:42.999
-It hasn't got, it's hard.
+It hasn't got... It's hard.
 
 00:22:43.000 --> 00:22:45.159
 It's like, really, there's a lot of work to go into it.
 
 00:22:45.160 --> 00:22:47.559
-So I think let's all just appreciate everyone who,
+So I think let's all just appreciate everyone
 
 00:22:47.560 --> 00:22:50.679
-who like contributes and makes all of this possible.
+who contributes and makes all of this possible.
 
 00:22:50.680 --> 00:22:52.159
 Cause it's, if you ever read
 
 00:22:52.160 --> 00:22:53.279
-the Emacs Develop mailing list,
+the emacs-devel mailing list,
 
 00:22:53.280 --> 00:22:55.479
 it's a lot of work, a lot of deep thinking,
@@ -1443,8 +1469,11 @@ all the libraries, all the LLM stuff we mentioned before.
 00:23:07.760 --> 00:23:09.839
 You're all doing such a fantastic job.
 
-00:23:09.840 --> 00:23:13.799
-It's exciting to be here. Thank you for your talk, Andrew.
+00:23:09.840 --> 00:23:12.323
+It's exciting to be here.
+
+00:23:12.324 --> 00:23:13.799
+[Corwin]: Thank you for your talk, Andrew.
 
 00:23:13.800 --> 00:23:16.159
 It's been just fascinating.
@@ -1455,6 +1484,8 @@ If you don't mind, I'd love to jump
 00:23:19.760 --> 00:23:23.879
 right over to the last question. OK, let's do that.
 
+NOTE Q: I have 45+ years editing, programming.   I'm not sure I can think about things without thinking of buffers, editors etc.   Is this a handicap/should we just have people with no experience with code learn to prompt?
+
 00:23:23.880 --> 00:23:27.599
 It says, I have 45 plus years editing programming.
 
@@ -1474,7 +1505,7 @@ Should we have people with no experience
 with code learning to prompt?
 
 00:23:35.320 --> 00:23:38.799
-Well, this is something I do have a strong opinion about.
+[Andrew]: Well, this is something I do have a strong opinion about.
 
 00:23:38.800 --> 00:23:42.159
 I feel like I do not want to see people that have no experience
@@ -1522,7 +1553,7 @@ Software engineering is making sure everything is scalable
 and all sorts of things that it's unlikely,
 
 00:24:19.600 --> 00:24:21.759
-I think, that an LM is going to get right.
+I think, that an LLM is going to get right.
 
 00:24:21.760 --> 00:24:25.279
 And I've seen a lot of bad cases
@@ -1531,13 +1562,13 @@ And I've seen a lot of bad cases
 where people who don't understand code
 
 00:24:27.200 --> 00:24:31.159
-are doing things and it's not working well
+are doing things and it's not working well,
 
 00:24:31.160 --> 00:24:33.239
 because they don't understand
 
 00:24:33.240 --> 00:24:35.279
-some of the some of the complexities
+some of the complexities
 
 00:24:35.280 --> 00:24:36.959
 or some of the concerns that that you might have
@@ -1581,11 +1612,8 @@ not 45 years, but a lot. And, and I think that it's those,
 00:25:13.080 --> 00:25:14.839
 those years of experience will only help you.
 
-00:25:14.840 --> 00:25:16.919
-And I think it's, it's, it's a great,
-
-00:25:16.920 --> 00:25:19.639
-it's great to kind of dip your toes in the water
+00:25:14.840 --> 00:25:19.639
+And I think it's great to dip your toes in the water
 
 00:25:19.640 --> 00:25:20.732
 and see what you can do.