Balancing Tech & Human Creativity • Kaiser, Greiler, Carpenter, Terhorst-North & Wardley • GOTO 2024
Welcome to "GOTO Unscripted." My name is Simon Wardley. I'm here with a panel of lovely people. Adele Carpenter, who came along and did some mapping, so a mapper as well. Sorry, I'm just a little biased. I'll throw that in there as well. Michaela Greiler, Susanne Kaiser, and Eddie, the shipboard computer. Hi there.
Hi there. God, I'm sorry, this is Dan North. So, we're here at GOTO in Amsterdam. This is the first time I've actually been to a conference in Europe for, oh gosh, five years. I think the last time was with Eddie, the shipboard computer, where you had an entire floor of a hotel as your room. They accidentally gave me an entire floor of a hotel. So, I checked into the hotel and it's just, you check into the hotel, and it's a really
quirky hotel called the Amrath in the city. And they said, "You're in room 501." I was like, "Okay." And they said, "No, you're in room 501." "How do I get there?" "You take the lift." So, I took the lift. The lift only goes to the fourth floor. Like, where's room 501? So, you come out to the fourth floor, you walk along a corridor, and then there's a separate staircase that goes up because it's a... It's a private thing. Well, because room 501 is the entire 5th floor. Wow. So, I went up this private staircase and you go in, it's like a sort of souk thing. So,
you go down a couple of steps, a big floor area, up a couple of steps, the other side, and then into, I won't call it a bedroom, it's a sleeping area, which is this vast four-poster bed. This thing had two balconies. So, you go through like a lounge area, a bunch of sofas, open onto a roof terrace where there's a huge table, you know, maybe 15, 20-seater tables out there. And then the other roof terrace. Because you kind of have too many roof terraces, right? There's a smaller roof terrace over here. And of course, I'm a speaker at a conference, and the first thing I think is, "We need a party in this room." So, we bought a bunch of people back, and yeah, and we partied in room 501. Cool. Thanks. It was, you just turned up to the hotel and it's
room 501, you're expecting a small room. You go up and there's a doorbell and it opens, and you're trying to think, "Where's his room?" And then suddenly it dawns on you, it's the entire floor. The whole thing is his room. Completely mad, but there we are. So, how are you finding the conference? Awesome so far. So, it's now the first
half of the first day. And I already have a lot of formal fear of missing out for all the talks that I could not attend at the same time. So, I'm looking forward to all the recordings. And now you are speaking as well, or you've spoken? I've spoken, so before lunch. Yeah, so I'm now relaxed.
Off duty. You've spoken as well? Yes. I'm also already on the relaxed side. There's a panel that I will be on tomorrow. But yeah, for today I'm done. And I'm just enjoying the company. I really enjoy connections. I haven't been to a conference for a long time. Meeting people is such a different experience,
so I'm very excited about that one. So, you're like me, it's been years since you've been to a sort of event. Unfortunately, yes. Adele, you came along to my...I did a workshop yesterday on mapping. You did. It was excellent. Thank you. Eleven out of 10, do recommend. Eleven out of 10. Fantastic. Thank
you very much. That's very kind. We didn't cover math... No math. ...or percentages. No, not at all. How have you enjoyed it? Well, a disclaimer, I'm on the program committee. Fantastic. Well, there we are. And I'm taking notes. Taking notes. Never invite Simon again. He's been quite useless. I'm very proud of the program that we have put
together. And I was in Susanne's talk earlier, and I thoroughly enjoyed that. She put a lot of content, including Wardley mapping into a 50-minute slot. And I saw another talk from another wonderful speaker, Jody from JetBrains. That was an excellent way to cut the hype of LLMs, which I think we could all use a healthy dose of. So, I thoroughly enjoyed that talk as well.
We just had a great keynote. So, I'm very, very proud of the program we put together. None of you... Good. ...have disappointed me yet. Sure. I promise I will do it tomorrow, but anyway, the... It all goes down with you tomorrow morning. Can always count on you, Simon Adaptive Systems through Wardley Mapping, DDD, and Team Topologies Absolutely. Well, what we can talk
now about LLMs, we can also talk about your talk. So, you had...well, I presume this was all about organizational design. Did you take everybody through the hands? Not only organizational design but also like, I call it arc, architecture for flow. So, it's kind of like a holistic view on three perspectives of the Brisbane strategy with Wardley mapping, of course. Then also software architecture and software design was domain-driven design and
team organization with team topologies. And bringing them together and also having a Wardley map as a conversation starter to generate a common understanding of the landscape you're operating in. And then also, I'm sorry, I'm reusing your Wardley map in totally different ways than you... I'm actually fine. ...might have originally attended or invented for. So, I totally... That's fine. ...I used Wardley maps in every...I also added layers. I have not presented it yet, only a little bit in this talk, but usually, I also added some additional layers to...
That's all right. ...going into the domain-driven design perspective from the problem space, starting with the user and the user needs. And then also diving into solution space where you come then with the components fulfilling the user needs directly and indirectly.
And then mapping them to evolution stages to also channel or to focus your development, your strategic investment in the core domain web. Also blending and domain-driven design... Nice. ...which comes in with the sub-domain types of core domain providing a competitive advantage. Supporting sub-domains and generic sub-domains where you can then go towards the right spectrum of award limit. More like, "Okay, can we use off-the-shelf products or all also the utility suppliers for those parts of our system that do not differentiate us?" And so it's like I combine everything. And then also...
Fantastic. ...bringing in also the team perspective. Also on the Wardley map and to try to drive some team ownership boundaries with domain-driven design as components. What teams can own what component on the value chain. And
also switching then from the external view to the internal product view to come also to the platform perspective was developing experience, also involving, like, product mindset and so on. So, it's a combination... Awesome. ...obviously, bringing them together with a goal to design and build adaptive social-technical systems. That's all. Look at all the trouble I've caused. I mean, that was fantastic. And you've got a book coming out as well. Yes. Yes. So, yeah, I announced it already. I'm sorry for the audience. I already announced it
I guess a couple of years ago and I was really naive, to be honest. So, it was my first book and it took so much time. But I'm also in a serious signature series of Vaughn Vernon’s Signature Series at Addison Wesley. So, it also takes, until you are the one that gets reviewed, it takes the time as well. But it's moving forward and let's cross fingers. I'm not making any promises when it's going to be published because there's always problems.
Right. But, yeah, I'm currently 12 years into my... Oh really, 12 years? ...first book. Amazing. And I've just decided, I actually have a strategy now. I didn't have a strategy for a long time. I was just kind of googling along. It just turned into...basically, it's got 40-something chapters because it's 40-something patents. My software, the stuff I've been doing for well over a decade.
And well, I'm working with a brilliant co-author now and what she's convinced me to do is, because the book is in chunks, and let's just take the team patterns chunks out and just do that as a book about patterns of effective teams, which was a talk I did GOTO Chicago a few years ago, kind of goes over most of those. We've got a couple more bits in there as well. And it is nearly done. I'm like, "I'm going to publish a book hopefully this year, right?" Wow. Awesome.
But it's only taken me 12 years and a massive kick up the backside and a brilliant co-author. So, you need those two things, I think. That's true. You see, I have a completely different plan in the fact that once your book is published, that'll be the 17th book with Wardley Map in it... Awesome.
...which will go onto my level library, which doesn't include my book because I haven't actually written it yet. And so... This is exactly what I do. This is my BDD strategy. That's like, "Why do you write a BDD book?" [I wrote a BDD book. John Ferguson Smart wrote a BDD book. Gaspar wrote...There's like some really,
really good BDD literature out there. I don't need to write one. Go and read those, they're brilliant. Rethinking Developer Experience: Tools, Collaboration, and AI Michaela, you were talking about, you spoke today as well. Yes.
And you were speaking about... Developer. ...DevEx, developer experience. Developer experience. Yes. Which is, I'm...Okay, so I'm gonna in my primitive sense, I mean, you are talking about the tool sets we use as developers, which are pretty much all identical, search, edit, navigation, and they're all appalling. Is that correct? I think that's in the essence what everybody understands around developer experience. And I think it also depends a lot on the productivity side, right? So, what a lot of companies...I mean, developer experience also
comes a lot from the companies that are developing software developer tooling, right? The developer experience is so much more, right? So, when we set out, I think around three years ago to really study developer experience and I did the ground theory study around that, and we really sat down and interviewed people and looked at the topics that emerged, right? The tooling is just a really tiny percentage. It's about the social-technical systems that we have, right? So, we actually identified 26 different factors around developer experience and we grouped them into 4 categories. One of the biggest is collaboration and culture, right? Like, how supportive are my teammates? How much knowledge sharing is going on? Do I get answers to my questions, right? A lot of things were about product management, right? So, it would be more developer-focused already, but it's about having clear goals, requirements, right? Can I do my job, actually? And then we had, obviously, the tooling side as well, right? But there was a lot about flow, about can I actually do the stuff without hitting obstacles, right? What about our, you know, feedback loops that we have? So, developer experience is so much more about tooling. I think it's...you know, developer experience, the terminology was a little bit used, like UX, user experience just for tools that are made for developers. But that's not the developer experience that I'm talking about.
That's not the research that we are looking at. This is really about how the team is doing, going away from how fast they are. Productivity metrics and measurements are really bad for knowledge workers, right? We're coming out of an industrial revolution time, right, where we measure how fast and how many things can we do? But it's not what we want people to do, right? We don't want them to write a lot of code that's, you know, buggy and we want to write them good code, innovative code, right? We want them to have creative solutions. I'm coming a lot from
the research side as well, right? So, there's this happiness, there was this productivity research, right, I was heavily involved in that. But there's also the happiness side, right, which is a little bit fluff, right? I know making it a little bit stronger, but a developer experience is not about that, right? Developer experience, and I think this is why it's so powerful, I call it doing your best work joyfully. It's really a work focus, right? It's really about thinking, the best work. So, quality work that you can do, but joyfully, right? So, you are getting back all this... How to make engineering joyful? Yes. But also doing really good work, right? And we mapped that back. So, what I presented
were some statistical analysis results that we got, right, where we really could show that there's a really strong correlation between the developer experience areas, dimension, right, the factors and outcomes like creativity, learning, innovation, right? Productivity actually is the outcome, right? The direct outcome from it, tech tap and all of that. But isn't that all going away? Because I was talking to a CIO a couple of months ago, and they're intending to get rid of all their developers because they're gonna have a prompt and they're going to be able to just type into it, you know, make my legacy system future proof and magically OpenAI will solve all their problems. Well, I tell you, my 8-year-old... Your 8-year-old. My 8-year-old taught himself to program in Python. Completely self-taught, auto-detect, with ChatGPT within three months, right? He built a web server. He built a compiler, a tiny compiler, a tiny web server, everything tiny. But it's working, right? So, I think AI is super, super powerful, right? But it's just a tool, right? And it's actually feeding off the information that we are feeding in, right? It's like, it's garbage in, garbage out. But also we have a lot of good information, a lot of good, like, code blocks,
for example, that it feeds off. So, actually, there comes working code out of it. And then if you have a person that steers it...and I think that's come back to the industrial revolution, right? It's just, we are even more knowledge workers now, right? Because all of that, things that we can actually put into pipelines, right, that we can automate. I think it's brilliant that it's...I mean, we have to be very careful and there are a lot of things that we...you know, we
should guide where it's going right? And on what information comes out and how we are using it. But I think I'm not scared at all for us developers. So, you don't see it as a replacement... No, not all. ...but as an assistant. Yes. Yeah. And we have to make sure that we keep it like that, right? Like, we have to keep it
within the boundaries of itself right there. Unfortunately, there's a lot of executives who think it is a replacement. Well, and this, I mean, my punt on this is with any technology, the successful technologies are the ones that help people do work better, not the ones that replace work. And I was saying earlier, kind of the architecture of the Toyota, the whole kind of lean production system stuff. You know, he talked about automation, which is automation with a human touch. And he says, you know, machines don't build cars, people build cars with machines.
That's true. And if you have better machines, you can build better cars. If you have machines that you can switch up really quickly, they call it a SMED, single-minute exchange of die. So, you could change the whole machine over in less than a minute. You can build lots of different kinds of cars, but you still need human
beings doing that. You know, calculators didn't get rid of accountants. They made accountants much less error-prone, right? It meant they could work better. Computers didn't replace administrators. It made admin much easier, you know, and the great tools are the tools that allow people to either do what they need to do better or to do normal things. And they...
Can I add something here? Sure. One of the factors of the effects, right, was engaging work. Can you do stimulating and engaging work? And it was very predictive for innovation, you know, for all the good outcomes that we actually want. And I think if AI takes over the boring task, like boiler-plate coat, right, the surroundings, and you can really focus on creativity, on innovation, on, you know, making sure that the tool actually gives you a good answer, right, and making sure, if the answer is even good and testing that, and, you know, I think this is amazing, right? I'm actually quite positive about it, but I obviously see also the problems that we have with it. But it's more from an ethical perspective, I think, where we have to be conscious, right? What kind of information do we feed off? What biases are there in our systems, right? And how do we use that? But with every powerful technology comes a lot of responsibility as well, right? So...
I agree with that. I think, you know, I look back at...and this is why, and I never think of feature parity as a product thing. It's all about experience. I remember for many years Eclipse was this big open source Java editor and IntelliJ was this big Java editor, and basically they were like feature parity right down to the detail. And they'd leapfrog each other a little bit. But, you know, you would find working in one, working in Eclipse, you could get stuff done, it was just a bit clunky. And working in IntelliJ, you would get the stuff done, it was just really obvious. And naturally, it just became an extension of you. You could, like, basically think Java into the machine. Do you know what I mean? And what they did
brilliantly was DevEx, was the user experience. What does it feel like to be using this tool? If you feel like the tool is on your side...And I guess, you know, when I'm building...like, I've been building trading systems and, like, traders do not want to use trading systems, they want to trade. That's what they wanna be doing. The trading system is in between them and trading, right? So, you know, if you can make the trading system invisible, you win.
That's true. If you can make the tool feel just so intuitive that it's not even there. So, there was, oh God, I can't remember the date. I think it was in 1972, Nicholas Negroponte published a paper, which was "Architecture by Yourself," which was all about when you design, the process of design is a conversation between two designers, sometimes within the head of a single person. And what's interesting about large language models is the system itself is becoming the designer. So, you are having a conversation with the system to build something. And that's the whole area of conversational programming.
But what's fascinating about that space is that a lot of the conversations we're having through things like Copilot are all text-based. And the problem with text alone, as in code, is it's stuck within a world of rules and syntax and style. So, is the code right and wrong? How have you styled...you know, how have you written that? But if you look at most engineering departments, the conversation occurs at two different places. One is the screen and the second conversation is on the whiteboard. And the whiteboard is about objects, relationships, and syntax. So, there's very, very different conversations going on whenever we're solving
a problem. And we've concentrated on one area, one specific, and using that as the sort of Copilot to help us. Now, one of my concerns about this is not only are we training it potentially in the wrong medium. Back in 2005, I introduced a structure, which is nowadays known as explorer as villages, town planners, so basically mapping out the environment, breaking it into small teams, small cells. And what we noticed is as the components evolved, the attitude you needed, so
you had skill sets like engineering, finance, but there are different attitudes. So, exploring, when you're exploring that engineering side is subtly different from sort of like the villager where you are turning things into a product to the town planners where you're industrializing, and you need three very different attitudes when building something. And they're all really important. ...I can confirm that there's also an ego element to this. Ok. So, I've always wanted to think of myself as, you know, the pioneer, the explorer, the, "I'm gonna go and have all these great ideas." It turns out I'm rubbish. I've had 2 great ideas in 30 years, 2. One of them is impressionable,
in 30 years. I'm really, really good at growing ideas. I'm really good at taking someone else's thing and growing it. It turns out I'm a brilliant villager and I kind of really fancied myself as an explorer and I just don't have an explorer gene in me. But I'm great at villaging. I'm like, I've made my peace with that by now. It's difficult. So, Susanne, I know you've expanded on this stuff so much. How do you find large language models and the idea of Copilots fitting into that?
I think so, it's also like when I see it from the team cognitive load perspective, I guess it could be. So, it's a great tool that, yeah, increases your efficiency in terms of, like, that you can use it like conversational programming, engineering, something like that. And also being your rubber duck when you do pair programming with your Copilot, for example. But also built on top of large language models in your own value chain that you can build upon, and then you can focus on your core domains or your core capabilities that you would like to focus on. So, it can reduce ourselves taking over or reducing potential team cognitive load. Coming back to the explorer, villager, and town planner, it's also my...So, my personal view,
it's a mix of mindset per each team. So, where one specific mindset is more prevalent, for example, like the one that tries to explore and discover new things, but they also have some of the other mindset maybe also because some areas needs to be stabilized and, for example, but also, like, the teams that own components in commodity, which are stable, standardized or more stable, more standardized. But they also have this town planner mindset to mature and optimize. But they also need to confront specifically when it comes to the large language model when it's new to their platforms, for example, they provide and how they can incorporate in their variety of platforms that they provide. But they also have to have a kind of, like,
explorer mindset as well, so that they enable those teams to respond in different ways at different situations. So, when something's new, it's not that the other team's exploring it, it could also be the platform team. Even though they are more on owning these stable components and the available chain, they also can say, "Okay, let's try this out with a little explorer mindset as well, and then see where we can also support the streamlined teams, imprompt engineering or something like that." So, that is, yeah, also that… Organizational Changes with AI and Tech Evolution An awful lot of change, an awful lot of, you know, this particular set of technologies impacting the way we think about things, the way we communicate things, the way we work with others, the way we structure. Adele, I mean, I would imagine conferences going forward, and you are gonna be bombarded by huge numbers of talk subscriptions, all generated by large language models. How do you distinguish between what is actually genuinely human and not? Secondly, of all these sorts of conversations, the whole sort of conversational programming to the organizational structure, fantastic event you've put on. What do you
think is really important going forward? Well, the GOTO program of GOTO is actually invite-only. So, we kind of dodge that whole delving into the... That's crafty. ...CFP landscape nonsense. Well, you still got material testing, right? You sort of know that the person you're contacting exists and isn't just a very clever construct... Important. ...of Wardley map somewhere. You make sure they turn their profile view. It's not Eddie, the shipboard computer. Exactly. We do tend to avoid that. But I do have friends that collect a lot of CFPs. And there was actually a screenshot that was going around that was hilarious where they had ordered all of the CFP proposals in alphabetical order.
And there was just a section that was in this ever-changing landscape and there was like 30. Oh, no. There's a great video from Dylan B., who's also well-known to the GOTO team. He did a video recently actually on that phenomenon of CFPs. And the change in words that have been used, a word like delve, for example, had a very low usage rate pre, you know, 2022 in CFPs. And now last year it's gone up orders of magnitude, the usage of that word. So, it's quite amusing to see. But yeah, personally...
So, you scrape out the word delve and that's like all of your LLM. Basically, yeah. Also colons in subjects or in talk titles is also a bit of a giveaway as well. I've got one of those. Me too. I just need to check some things. Well, that brings us all the way back to Eddie,
the shipboard computer. So, I'm gonna say... I'm busy right now, I'm trying to make a cup of tea. A cup of tea. Right. So, all I can say is thank you ever so much. Thank you for joining us for
this very "Unscripted GOTO talk" and you've been absolute stars. Thank you.
2024-11-15 00:39
