I don't think I've figured out the best way to go about giving a conference talk. Pretty comfortable that I can give a very decent conference talk and I get good feedback on my conference talks, but the process to get there is terribly inconvenient and mildly stressful as well as just like time-consuming as I'll get out, but I am preparing to give several talks at the moment. So I think I'll stick to my current approach because I know it can deliver results, but I also think I need to practice something that is less demanding in prep. How do you deal with public speaking or do you deal with public speaking at all? There's proof of me speaking publicly on YouTube. There's probably more proofs, but this can be found by just YouTube and on my name and I talk about Sveerok, how the organization is, the structure of the organization, and this talk was unrehearsed. I don't think I had slept for very long and it was all kind of ad hoc, and it never really should have been filmed and put on YouTube, but... There was a gun spiked in views, because I'm going to look at it very good, but there was a guy there who brought his phone and just filmed it all and cut it up and did something kind of good with it, and he seemed to think that it was good enough for him to spend lots of time editing it, so that it's good enough for me to have on the internet. Cool. Yeah, and it's over 10 years old by now, so my usual way of doing public speaking is ad hoc and just winging it and talking about something I'm very passionate and know a lot about, so that's what I usually do. I had done some more planned speeches too or talks when I presented my thesis work for instance, that was... Then we used beamer, I think the name is beamer, you write your slides in La Pech and you can write all the good notes and stuff that's very useful, but then there was hardly any time at all for anything, so we simply read it, read the text verbatim, so I think that's kind of a cheat code to be able to read a text and not sound like you're reading a text, but... Yeah, that helps. Yeah, but it's not at all as good as just knowing the stuff by heart. One of the worst things to see is someone giving a talk and to see on their face that they didn't at all expect that slide there. I've had that for fairly well prepared talks sometimes, well, for the keynote I gave at GIG City Elixir, I've been sort of working on the presentation a fair bit and I've rehearsed it a fair bit, but there were slides which I'd just stopped reading while giving them, so I was completely unprepared for my audience laughing at certain points, because the way I use my slides is usually a blunt instrument where I hit people over the head with them while I explain something, or rather, I don't necessarily think, so I think the slides should match roughly to what I'm saying, but I don't think the slides should be exactly the words I'm saying. Yeah, I agree. So that they reinforce each other and can feed off of each other and I can bounce things around a bit, and I can do a word gag on the slides that won't work in person or I can make a joke that requires kind of voice of timing while I'm speaking, that might read straight in text, yeah, so that actually caught me out of words at one point, I don't have time to read the slide to figure out why you all are laughing, I have a lot of stuff to cover. It's like that meme, it's a talk, it's not meant to be fun. Yeah, got lots of work to do. Yes, pay attention now. Yeah, and I think there's also something to be said for giving talks that are a little bit more sort of down the middle in terms of, well, the combination of presentation style, preparation, and entertainment value versus content and all that, like, I've seen talks that I could not listen to, they were contentful, but were not entertaining enough to keep me alive when I was in an intense conference. I've seen speakers that can really capture you and then deliver nothing, but it was fun, right? Well, up until up to a point, like most of the talks that are really fun and memorable are usually also like they have some content to them or they have, they're presenting you with an idea that's interesting and like trying to combine presentation style, preparation and presentation style and preparation go kind of hand in hand, but also like entertainment value and content, like finding a good mix of those, that's, that's hard, like, or at least it's, it's a decent bit of work. I'm writing my talk for Berlin now and I have a lot of what the contents should be because I'm talking about nerves and talking about nerves hub. I'm going to talk a little bit about nerves cloud, the startup we're doing and there's a bunch of features I can cover and there's a bunch of functionality and like I have so much to show, but I have to make it interesting beyond just like features blow by blow. We built this because I don't care, do care to do that. I find it dull to do because the features are very good. I could probably get away with it and just like, okay, a little bit of flair and a lot of features actually makes for pretty decent talk. I think that's typically the style of like Jose and Chris McCord in the lecture community like they're good speakers. So they pull it off really well. Jose might not always do like a features talk because like Elixir doesn't change so rapidly that he has to. He talks a fair bit about bigger picture, but the last Chris McCord talk I saw in person was just like cool feature rundown and with kind of a historical perspective and a bit of story to it, but it was good. Like it was a good talk and it's what a lot of developers want. It's like, tell me cool shit I can do. But like my first talk at ElixirConf was kind of a stunt in some ways where it was a live demo that, well, I controlled my slides with my voice using the whisper AI model while, yeah, my slides was a demo project for what I was talking about and the topic was kind of developer creativity and why Elixir is so good for that. But then there's like, I can't, I won't do the same thing again. I will do something different. So, and I think the way you present and the contents of the presentation and your level of preparation and all of that kind of feeds together into making a memorable or good talk or good presentation. Fundamentally, I'm not sure people truly like my presentations because there's always an imposter syndrome lurking about and in some ways like I know more people will tell me it was good than I will ever hear on blunt, honest feedback. With developers, you can usually get some blunt and honest feedback, but I don't trust that I've heard all the opinions about my ability to do presentations, but I have enough numbers to kind of back up that I probably don't care. And I feel okay about the way I do it. And that's the most important part, right? Or is it? I'm not sure. So I, I care a lot about sort of reaching the audience with my message as I intended it. I don't need to hit everyone. I need to hit the people that get me, that get my ideas. It's fine if it flies past a few people, but I do want it's like previous out of the dev, I gave a presentation on like live view and why that can be your entire new development stacked essentially, but it was a, it was a quick run through a lecture of Phoenix and a bunch of stuff around the ecosystem and kind of giving some quick kicks to larger ecosystem along the way. And for example, I'd be a little bit bummed if a lot of people came up to me and said that I was really rude about other ecosystems or whatever, just because then they probably didn't register or didn't care that it was all presented in a fairly loose and joking fashion. It was like, at the same time, like I've done enough public stuff that I don't expect my intent to reach exactly everyone, because there are people that can't read intent or read that type, sort of that signal layer at all. So you will miss some people whenever you use like sarcasm or joke or whatever or reference culture in any way. So I don't have that as a kind of primary concern, but I also don't want to, if you really screw up your, the way you're joking, you can come off as really arrogant, which is, I'd prefer not your worst case, get thrown out of the conference. Yeah, I would say that's an extreme. Yeah, there are only a handful of those examples. So don't do that, please. Yeah, I'll try not to. Oh, very good. But what's the, I think I read in one of your newsletter, so something how you prepare for talks. But that was a long time ago. So I'm just going to guess how you prepare for talks. Okay. Yeah. Yeah. So I think that you sit down and then you write about four, four weeks straight, like just a smaller book. Then you go through this book and turn it into slides. And then you start cutting stuff while you rehearse this and hardly sleep for about six months. And then you have a talk. The timelines are kind of weird, but mostly accurate. Cool. I've been doing software estimation now for a while. So my timelines are weird. Yeah, I'm sorry. No, but there's a lot of writing. Like I'm so comfortable with writing. So I capture my ideas in text first, which is generally good, I think for me. But one challenge is that when I then start writing out, well, as a talk, it's more specific. I start to write out the wording I want to use, which I don't have to rehearse. Oh no. And remember, because I don't really want to read from a script, but like the better rehearse I get the more I also diverge from what I wrote. Like I can stick close to it, but I don't feel like I have to do it verbatim. But but yeah, I start out with writing until I have enough of a talk to start rehearsing it. And then I start rehearsing it as a way to iterate on the formulations and getting a sense of the length, getting a sense of if I need to put more stuff in, take something out. And as that happens, I also build up the slides. And depending on how I'm doing the slides, that will vary. Actually, the voice control was kind of helpful, because it was easier to remember the sections, because I had kind of code words that I used to advance the next slide, because like, oh, this slide ends this way specifically so that I can move to the next one. But yeah, the rehearsal thing, like I tried to rehearse morning and evening once every morning, once every evening, when I reached a point where I can rehearse. And then the point of that is to do it for so long that I get bored of it, that I get really, really bored with the text itself, with the contents itself. If I can give this thing into my sleep, and I'm sick of the material, then I'm very well suited for giving it and I'll be, I'll be happy with that. I usually don't get quite that far. I think I was pretty much there for my first talk in Lisbon, the lively membrane one. But other talks, I don't feel like I quite have gotten there. Usually I end up mildly ill at some point during the prep, and then then I lose a bunch of time, where there's other things. Yeah, but let's see, I gave a talk at Nervskow, just before a gig city elixir, like they, they were attached to each other, those two conferences. And that one, I didn't rehearse significantly, because I knew I wouldn't have time. Like I had a keynote to deliver at gig city, I could not do a talk where, where I needed to rehearse for Nervskow as well. But I wrote up something I thought was kind of fun, like a fun romp through some security work I'd done, and a lot of kind of jackshaving diversions. And then I, that one, I read a lot of my notes for. So, but it was also a pretty decent setup for that, because you had, you could see your slides, you could see your speaker notes in a decent way at that venue. So it was a bit like coming in teleprompter. Yeah, a little bit. I don't like how that breaks my, my ability to look at the audience. But it's not like the audience will always notice, because they are usually looking at your slide screens, the screen showing your slides. And as long as you sort of regularly sweep across the audience, they will know that you're not just reading. And since that talk also had a decent bit of like interactivity in terms of me shouting out people in the community, and a lot of them were in the room, and sort of a little bit of back and forth, a little bit of that fun stuff that I think it worked out pretty well, right? Yeah. And so that one was a lot less prep. So I could write, write it, edit it, like read through it a few times for both for practice and just for knowing the contents of it and figuring out where it's weak and where it needs more or whatever. But yeah, so now I'm prepping my big talk for code beam Berlin. It's kind of cool, because I, the concept is very cool. So now I have to do a lot of cool shit with it, or I'm wasting a bunch of community effort. I don't know if you've seen what I'm up to there. I might be completely wrong here. Is it the James Bond villain headquarter's map of the world? Yes. Yes, it is that map. Oh, yeah. That's my colleague implemented that kind of prototype map for nerfs hub recently. I'll have to yell in for that. Oh, nice. I currently have 104 devices that have been donated by the community to my talk. So good. So unfortunately, the filtering is not so good yet that I can just get counts of the various types and stuff. But there's a lot of Raspberry Pi 3s, a lot of Raspberry Pi 4s, a decent bunch of Raspberry Pi 0s, some 5s, some 2s, some 1s, a couple of grisps, a bit of mango pie. And then there's a whole bunch, I think, of smart rent hubs that John Karstens had and decided he could connect. So they made essentially a training version of their hardware that they use for their thermostats when they were doing an earlier nerfs-count training. So that's kind of cool. So we brought up a bunch of those as boards you can use or as boards from my experiment. I would love to have a bucket or gram or histogram or what's the name for that would just like here are all the machines, the number of them. I need to check how many original Raspberry Pi's do I have up. That is not a strong machine and also, okay, so three of them, they're all online impressively. The original Raspberry Pi did not have Wi-Fi or ethernet. How does USB dongle, baby? Or like, you are serial or something. Malamodum, we need more of those. I'm blinking lights. Yeah, maybe you could hack something over the composite video. Yeah, that should be faster. I have one of those, an OG Raspberry Pi somewhere among my electronics. I need to dig it up and do something stupid with it. I think it might be slower than the Raspberry Pi Pico. Yeah, I think so too. Hmm. I'm not sure about that. But the Pico can still run Linux, right? No. Well, you can probably get Linux to run on it, but no, not as much. It doesn't have the memory blah thing. Yeah, memory management unit. Exactly. Yes. Hmm. Interesting. Yeah. So much cool hardware. Okay, so you'll do some here. Watch the world while I press buttons thing. There will definitely be some demo stuff there. I have some plans for making this Armada do some kind of work, some kind of coordinated work. But yeah, I haven't pinned down the details. It depends on how much stuff I can get working. So I was poking at like, oh, can I get the whisper AI model onto these things? And can I get the MP3 library I need? It's probably doable, but I had a hell of a time with my first attempt. So I might end up doing something significantly simpler. The point is not really to do a bunch of awesome AI, because they're all CPU machines. Like maybe someone has attached an accelerator. Actually, I think at least one has, but it's not like I know and can use it. It's very hard to sort of ask questions of all of these machines at once at the moment. And this is also something I would like to make better. But yeah, I have roughly 100 devices. I don't want to connect them all into a cluster. I really, really do, but I shouldn't. Because an airline cluster is not to be trusted, or well, an airline cluster has to be trusted. So if I connected them all, which I absolutely could figure out a way to do, like wire guard and stuff, then I would have the problem of everyone having remote code execution on everyone else's machine. And that's not what they signed up for. They signed up to give me remote code execution on everyone's machine. Yes, yes. Also, you get the exciting m squared messages thing, right? Oh, yeah, yeah. I'd probably stretch the airline mesh approach a little bit, but it could work. I guess this pie zeros would struggle. Maybe the original pies as well. Yeah, yeah, even like my co founder, Yash was like, can I get access to to the product account thing? So I can actually look at them and figure out some stuff. I'm like, not really, because I can't give you access without giving you too much access. And that's not what I asked for. So like, I have to be fairly principled. When it's like, at least 50 people's networks. Yeah, maybe do that in the good way. And he asked very kindly and he backed off very kindly. So he's a good egg. It's just like, Oh, our product has actual items in it right now. I want to I want to see it. I want to see where it breaks. I want to see where I can fix it. Yeah. Yeah, I get it. Yeah, beyond that, I have some ideas about what I want to do demo wise. And then I have a few ideas about presentation style. And I have an idea I really want to do around how to do the slides, but I won't talk about that too much yet. Because I'm for one thing, it might not work very well. And also, I might not have enough time and resources to do it. And also, it's better as a surprise. If I do it. That's, that's a great hat trick of. Oh, shit, or something. Yeah, all my reasons. Yeah, good ones. So, so speaking of slides, are you in the more slides are better or less slides are better camp? Them correct amount of slides is better. Cool. What's the correct amount of slides? Depends on your talk, right? Yeah. So if you're, if you have awesome illustrations that really bring your point forward or something cool that animates in a meaningful way or you're really talking kind of big picture and just want the word delightful on the screen for 10 minutes, like that can be the right thing to do. But if you're also like rapid fire going through your awesome, awesome project and like showing every nook and cranny of what it can do. Or if you have type of slides where you reveal a bullet at a time and you just go clicky, clicky, clicky, clicky, clicky. That's also a style choice. With that style choice, you might end up with like 500 slides, 2,000 slides. You never know. And I don't really care which, which approach you take. I, for my other dev talk last time, I know I did the kind of click to reveal for comedic timing. I used a lot of emoji for icons. It was really annoying because my emoji don't exist across. Well, across the slide software and yeah. Yeah, very rough. Yeah, that's a good reason to go for the printed PDF or bitmap PDF thing. Because then you don't get lost in the different software capabilities, operating systems and whatnot. Yeah, I think my new approach will prevent that. Cool. It will all be very rasterized if I do that approach. But then I'm also giving a talk. So the talking Berlin is in the middle of October. I'm going to a conference in September as well. I'm going to open source summit mostly for the embedded Linux conference inside of it. But I'm not speaking there, which is going to be interesting. I was like, oh, I'm absolutely just a visitor at this conference. That's going to be weird because I haven't done that in a few conferences now. Like I always get involved somewhere. I always end up doing something pretty much. Yeah. Yeah. So I'm doing that. And then there's a middle of October code beam Berlin and we're doing a workshop, a nurse workshop the day before the code beam Berlin conference, which is going to be super cool. Yeah. And then in November, there's a dev. And they had had some speaker drop out of their DevOps and cloud track. And I was I was poking around trying to get some functional programming people involved without a dev. And then I ended up touching the live wire and got considered for forgiving a talk. It was like, hey, can you just submit a suggestion and we because we need one? Yes. Yes, I definitely can. What I'm trying to do right now with nerves, a nurse cloud and all that is literally just blast attention onto nerves. And I can absolutely push the angle of sort of the DevOps and cloud, primarily like DevOps and iteration workflow and how nerves brings that to embedded Linux because it does. It has a lot of and and that part has a lot of layers to it. Yeah. And it's also novel and different and interesting enough for people that just do like online websites or services online, because embedded different world, like you're going to be very upset if you accidentally break a thousand units placed around the African content. It's enough breaking one unit that's placed awkwardly behind a wall. It's I had my fair share of swearing at that when the guy who was had to change the SD card had to crawl under behind and do some acrobatics to change it. At a certain point, one could say that it's just another brick in the wall. [Music] [BLANK_AUDIO]
