Shownotes

Title: The Future of Software Efficiency, Distributed Systems, and Face-to-Face Connections in Tech. In this first episode of Asynchronous and Unreliable, Anne Currie is joined by Sara Bergman, a Microsoft software engineer and co-author of Building Green Software. They explore the nuances of code versus operational efficiency, the evolving landscape of hardware and software, and the importance of face-to-face interactions.

Key Topics:

• The balance between code efficiency and operational efficiency in sustainable software development

• How AI is increasingly capable of optimizing code performance, including rewriting for efficiency

• The shift from hardware solutions to smarter software and operational practices for sustainability

• The strategic use of multi-tenancy and hardware flexibility to enhance efficiency and sustainability

• The significance of face-to-face conferences, hallway track networking, and speaking opportunities for professional growth

• The impact of transitioning from traditional hardware dependence to chip-agnostic, multi-platform code

• The industry's response to energy constraints, energy transition, and the potential environmental benefits of energy-aware design

• The power of community, feedback, and continuous learning through conferences and meetups

• The evolving role and perception of tech investments in a changing economy

Resources & Links:

• Building Green Software by Anne Currie and Sara Bergman and Sarah Hsu

• The Cloud Native Attitude by Anne Currie with Jamie Dobson

• Green Software Maturity Matrix (Online tool from the Green Software Foundation)

• Ampere's ARM Chips

• AWS Graviton

• Microsoft Azure

• Rust Programming Language

• Environment Variables podcast with Anne Currie & Sean Varley of Ampere

Find Anne at:

• LinkedIn

• Bluesky

Find Sara at:

• LinkedIn

• Bluesky

This episode emphasizes that success in tech sustainability involves not only technical innovation but also community engagement, face-to-face interactions, and continual learning. With tools advancing and energy concerns growing, the right operational practices and hardware flexibility are more vital than ever for a greener future.

Transcript

Anne Currie (00:00)

Hello and welcome to asynchronous and unreliable, a new weekly podcast where we discuss the most interesting ideas and concepts in tech and how they might help us in our day-to-day jobs or even our day-to-day lives. I'm your host, Currie, co-author of Building Green Software, the Cloud Native Attitude and author of the Science Fiction Panopticon series. And for our guest today, who could be more appropriate than my friend and co-author on Building Green Software, Sara Bergman. So Sara, do you want to introduce yourself?

Sara Bergman (00:30)

Yes, hi. Thank you so much for having me on. I am so excited to be here with you. It's always a pleasure and excited to be on your podcast. What a privilege. So yes, I also was one of the co-authors, one of the Sarahs the Sara of the Sarahs of building green software. As my regular job, I am a software engineer at Microsoft where I work with utilization and efficiency of CPUs on cloud scale. So for our internal workloads really. What else is important to me? I used to say I'm a Swede living in Norway, but now I live in Sweden. So I'm just a Swede in Sweden, which is kind of boring, but there you go.

Anne Currie (01:11)

it's such a pleasure to have you on the podcast. we're on podcasts quite often together, we're on the Green Software Foundation's Environment Variables podcast quite often, because obviously we wrote Building Green Software. And I always enjoy it. a lot of the reason for my doing this podcast is to have more conversations with with you and with Sarah and with other interesting people in tech, find out what's going on. Because my fundamental reason for starting this podcast is I'm nosy. I like to understand what's going on in the tech industry and why. And it's really useful to have people come on and tell me in their own words what's happening. And you're a really good candidate for that for two reasons.

One is that we have a shared interest in efficiency. You're working on code efficiency now and operational efficiency. And that's kind what we're going to talk about in this episode today. But the second reason why you're a great guest for the podcast is less obvious because people aren't going to know unless they know all about us, which is that we first met in person at the hallway track of a conference, QCon London quite a few years ago now.

Sara Bergman (02:19)

2022 or 2021?

Anne Currie (02:23)

Was it basically the first one after things started reopening?

Sara Bergman (02:26)

Yeah, 2022, could that be right? 

Anne Currie (02:30)

I think it was it was yeah and we did at that point we had no idea that we were going to write a book together you were just coming over to London to speak.

Sara Bergman (02:39)

No, but you were the one who got me there. I owe a lot of my career to you Anne.

Anne Currie (02:47)

I am a busybody. There's no doubt about that. I kind of did want to meet you in person because it does make a big difference. And conferences are really good way to get somebody else especially if you you get someone to come and speak at a conference, the conference will pay for you to come and then then then you can get people to come from across the globe to meet you without you having to travel very far, which I like. yes, so we met at a conference. And one of the things I want to talk about today is conferences and meetups and why it's actually still valuable for people to meet face to face, which seems like a strange obsession for me to be promulgating while I'm doing my podcast, which is all about not meeting face

Sara Bergman (03:35)

Not meeting to his face, yeah. At least we're looking at each other.

Anne Currie (03:36)

That is true. And it is nice because frankly, it's a heck of a lot of effort. Well, for me and for you at the moment. So there's a reason why I called this podcast asynchronous and unreliable because you can't always travel to see people. You know, you can't. I have multiple sclerosis. So I'm not always super well, which means that I can't always travel all over the globe seeing people, much as I would like to. Plus, I don't have limitless money to do that. And Sara, you are at the moment, although normally you're like a super healthy Swede, skiing everywhere, doing all the Swedish things, you are heavily pregnant. you can't.

Sara Bergman (04:21)

It's kind of trouble, which is not a new situation. Well, this is my second pregnancy, so I'm used to it for the first. But the first was very different, I have to say, for people who maybe had multiple pregnancies. The first one, yeah, we wrote the book when I was pregnant. I did my last conference at eight months pregnant. I did my last podcast, like four days before I gave birth. And this one has been completely different. I've been almost homebound. So, yeah.

I feel like I'm very much the unreliable part in our relationship right now. Like our relationship is mostly asynchronous anyway, because we're not in the same country, but right now I am definitely holding up to the unreliable part. So I do love the name.

Anne Currie (05:00)

Well, think it's good. with distributed systems, one of the things I like is you, get things that are more reliable by leaning into the fact that they're not reliable. But if you plan in advance for asynchronicity and unreliability, then you can cope with those and get really resilient system. So at the moment, it's possible that this will be the first ever podcast of asynchronous and unreliable, but I was going to not say that at all during the podcast, so that I could not have it be the first episode. my aim to cope with your unreliability and my unreliability was to buffer off a lot of podcasts, the classic distributed way of making sure that you maintain a weekly cadence. 

Sara Bergman (05:48)

By doing stuff in advance. feel like that's a life hack with so many, many things. It's like meal prepping or like laundry or anything that makes life easy. Do it in bulk.

Anne Currie (05:58)

Yeah, I do think, you know, distributed systems as a life philosophy is a good one

Sara Bergman (06:05)

It's good one. Yeah. Awesome.

Anne Currie (06:07)

But interestingly, so I love distributed systems and I love, I love efficiency, love code efficiency, and I love operational efficiency. So we should kind of park our why distributed systems is a life philosophy for later when we retire and running a lifestyle businesses. You know, calmness for software engineers based on application of distributed systems principles. Half the podcasts I listen to are selling some kind of calmness routine. Anyway.

Sara Bergman (06:40)

Is that a sign Anne? We can do retreats. That will be fun.

Anne Currie (06:43)

It would be yeah, distributed system retreats. Where you you know, butter all your sandwiches in advance. But that's not what we're doing. We're not doing distributed systems retreats We're actually today talking about efficiency because at the moment your specialist subject is that you work at Microsoft on efficiency, something that we kind of bonded over right from day one, the first time that we met over a zoom call and then met in person, because it's it's a really interesting subject. Particularly when it comes to code efficiency versus operational efficiency. And what you're working on at the moment, what you want to be talking about today, what you're working on at Microsoft is about write code once run anywhere, in terms of not relying on a particular chip. So not always thinking you're to have one chip in mind, being able to move from chip to chip, which I find interesting because it's anti what you would always traditionally have said to be the case for code efficiency because code efficiency is all about optimizing. Really super, super hyper code efficiency when you're going for really, really high efficient applications, you want to tune them particular chips, but your argument is it's more green to not tune for a particular chip. So tell us all about that.

Sara Bergman (08:19)

Yes. And I will say, I will say there's not like, there isn't the ideal answer for every scenario here. And I think that's also what makes it endlessly interesting to discuss because there are like corner cases, are groups of scenarios and blah, blah, blah. And I think what we have encountered a lot and [our co author] Sarah Hsu also encounters a lot is that when you talk about software, green software at a conference, the first question that always comes up is, what about code efficiency? And it makes sense because coding is very much our craft, it's something that a lot of people take pride in and I don't ever want to take that away. I think you should always write efficient code, like there's no point in not doing it. And for the hyper, hyper performant scenarios where every line of code matters, of course, that's what you optimize for. You don't optimize for operational efficiency. But what's interesting working at such a large company, as Microsoft is, we are both a cloud provider, but we have a lot of our own software, which I hope most people know or have used even. And we have great flexibility over how we place those workloads because we own them. And we also have full insight into it. Like for us, it feels like on-prem, I guess, even though it's still the cloud, because we own it. And that's the case, of course, because a lot of people still have on-prem or hybrid, of course. So yeah, for those workloads that aren't those hyper, hyper performant not that they're not performant, of course, performance is always important, but there are still classes that are tiers of that. Those are the ones we target. Yeah. And I think we had a job posting open actually for my team and then it got picked up. We think like more than we expected, we didn't expect that to be so much, but then someone deduced from that that we will be rewriting all of Windows for ARM. And that's not at all what we're doing. 

So ARM is a big trend across the industry. Some have been doing it for a long time. I think AWS launched their Graviton in 2018, which is eight years ago. And we have it almost everywhere now. Google now has their own ARM chip. Azure has our own, a bunch of smaller clouds also. Ampere chips are very popular. A lot of the, I say smaller, they're still big, but they're not as big as the big three. So that's really also a trend coming across the industry. So think we're going to see it more and more..

Anne Currie (11:07)

So the thing I find quite interesting about it is it exemplifies a big battle that we constantly come up against. Software engineers at conferences keep saying, can I rewrite my code in Rust or C to make it more efficient? And we always say, look, first start with operational efficiency. So it's all about battle between code efficiency and operational efficiency.

Code efficiency has loads and loads of potential. If you rewrite your code in Rust or c and you really tune it for particular chipsets, you might get 10, 100, 1000, 100,000. I've even heard of a million times improvement in efficiency. if you do that. But it's really, really hard to do. It's really hard to do.

Sara Bergman (11:57)

Very expensive in developer costs.

Anne Currie (12:00)

Yeah, it is. Operational efficiency is much more low hanging fruit. So there a doubling in your efficiency is achievable by almost anybody with quite a small amount of effort. But with a bit of effort, you'll get 10 times improvement, which on the surface of it fades into insignificance compared to the 100,000 times comes from code. But is relativly easy. I would say that getting up to that level of operational efficiency is just achieving modern operational competence.

Sara Bergman (12:37)

Yeah, for sure. Because you have to also keep in mind where people are starting from. Like a lot of people are still new to, like, a cloud scenario, that might not be something they are an expert in. You have written all about that in The Cloud Native Attitude book. And most people don't have that cloud native attitude just yet. And they're coming onto it. 

Another piece of work which you started was the green software maturity matrix. If listeners haven't heard of a maturity matrix, it's kind of set of categories and a set of stages and you can like, based on a set of criteria, identify where am I in the progress of in this case, like becoming a green software business. And you can be quite far back in some and quite advanced in other categories.

Yeah. And I think a lot of people are further back than they perhaps realize or want to, or need to be.

Anne Currie (13:40)

Yes, so the reason I'm saying this is a battle between code efficiency and operational efficiency: the idea of "green is write code and run it on lots of different chips", is that that helps with operational efficiency but it doesn't help with code efficiency. So if you're kind of going, if you're the kind of person who's leaping straight to code efficiency, saying I want that hundred thousand times improvement, then you'll go, well, why would I start here? Because that doesn't get me there. But as you say, I think you always have to start with operational efficiency. And if for most enterprises, the argument that we made in in building green software and the maturity matrix is that for most enterprises, you really only ever get to the top of being really good operationally, and you leave code efficiency for people who have the money to invest in code efficiency like Microsoft, like Amazon, or whatever, because it's so expensive and difficult to do.

You'll have to look impassive at this point, because you're not allowed to, as a Microsoft employee, talk about AI, but I will make a comment and you will have to say, "you may say that I could not possibly comment". I think that code efficiency is something that AI will get very good at. Because it's mostly just about rewriting code and it's quite measurable, you just measure off performance. It fits in quite well with AI but it's very difficult for humans to do because it's very costly, takes time and needs all kinds of specialist skills. 

Sara Bergman (15:23)

I just want to say that will, for me, be one of my favorite usages of AI because then we could have it all, you know? The cost side of AI, which we can give for another podcast, but yeah.

Anne Currie (15:33)

Yeah, yeah, I've got another podcast coming where I'm going to talk a little about how much more inefficient code is than it could be because it's written by and maintained by humans. You know, it's like, it is close to a hundred thousand times less efficient than it could be if humans were taken out of the equation. But so I've got a few folks coming on to talk about that later. 

Sara Bergman (15:57)

I'm looking forward to those episodes.

Anne Currie (16:02)

so, the future of code efficiency, and we did talk about this in building green software, the future of code efficiency is very hard to predict. And so I wouldn't at the moment start my career in a code efficiency but I think that operational efficiency is a more human skill. And it will be around for decades. So getting good at operational efficiency is just being good at ops. It's using the tools as they were meant to be used, using the platforms as they were meant to be used. It's auto scaling. It's light switch ops. I've got I've got high Holly Cummins coming on later in the series to talk about automatically turning things off when you're not using them. It's all the good things that give you security and resilience and performance and cost savings without the incredible difficulty of rewriting code in C or RUST.

Sara Bergman (17:03)

Yeah.

Exactly. And you can keep delivering a business value as you do it, which I think also it's interesting because I haven't been around in this industry for so long because I am hair flip quite young. I can't say that for many more years.

Anne Currie (17:19)

I've got 20 years on you, so you can say it for quite a long time.

Sara Bergman (17:30)

But I think there was a time where software was not cheap, but it was considered cheap. Holly [Cummmins], again, had a great talk on this in Copenhagen late last year on being in a ZERP zero interest rate period and how that has led to massive investment in tech, which we're forever grateful for because that has led to a lot of cool stuff for us as tech people and also tech users. But now that's changing, right? The economy is changing and we have to be more mindful of money, I think in all industries. But think our industry is less used to it, perhaps compared to a manufacturing industry where you're always thinking about budgeting. So that's another thing that operations is very good at because often the cheapest thing is also the greenest thing.

Anne Currie (18:26)

Yeah, yeah.

Yeah, we both are massive proponents of using cost as a proxy for for emissions. Yeah, and it doesn't take that much focus. I think the problem is, as you say, that we've been through a zero interest period. But because, as we discussed, I'm 20 years older than you, I was around in the first couple of decades of the tech industry, where we didn't have, again, where we were quite focused on money. And we were quite efficient then, we wrote really efficient code and we were much more operationally efficient or really careful about not putting in a single extra machine into your data center. That was painful. You know, I remember one of the my jobs 20 odd years ago in retail, we had our own data center and the problem was plugs. You know, it was like, we do not have any more plugs so you cannot write any more software. Unless it's efficient enough to squeeze onto an existing machine. So the world has changed massively. And we have become quite lazy and profligate. But I don't like to say that in a way that's that makes software engineers sound lazy profligate as even though I have now But what I have realized and noticed recently is that people have lost understanding of how powerful software is. They don't know how much you can do. And that's, that's spread entirely through businesses. They think problems have to be solved with hardware. So you know, even software engineers think problems have to be solved with hardware. They don't know how many problems can be solved with software. But it is hard to do. 

I want them to solve problems with better operations. I think that's a no brainer. That's kind of like, all enterprises have to do that. But then some enterprises like Microsoft also have to work out how to solve problems with software, which is harder, but more powerful. So it's an interesting one, isn't it?

Sara Bergman (20:40)

It is fun. Yeah. I think that's also what I really enjoy with my job right now, because we have to figure it out. So, we do have Azure. We have like capacity that we sell and that is used by customers. And then we have our own fleet, but we can be quite flexible. Like, of course, within the local laws and regulations and there, especially in Europe, there a lot of things we have to take into consideration, but we still have a lot of flexibility and insights into how we place that and how we kind of eat up some of the, let's call it waste. I'm not a great fan of that word, but unused compute or like scrappy compute or yeah, like have more fungibility, I guess is a good word where we can see like in one region, we maybe have a lot of customer demands and then in another one we have less. Great. Then it makes no sense to run our internal workloads in the region where we're almost at capacity with sellable computers, move it to a different one and things like that. That is a very interesting thing to look at at scale because I'm always like a scale person. That's where I think the most important and interesting problems come from. I imagine if you have your own data center, you get the same. Maybe not on like a regional level, maybe that's on, like you said, on a machine level. Okay, but that machine is full. Can we fit something in here? Do they have alternating usage pattern? Maybe we have a batch runner that we can run at night and then we have the ordering system, which is mostly used in the daytime. I don't know, but things like that.

Anne Currie (22:14)

Yeah, we talk about that in building new software. So that kind of multitenant thing is the benefit of the cloud. Cloud is an interesting one. I'm going to say nice things about the cloud, but I'm going to introduce them with the problem. The problem with the cloud isn't really with the cloud, it is with the customers of the cloud. But it is kind of also with the salesman of the cloud, who sell the cloud and don't make it clear that you have to move into the cloud and then you have to use the cloud. 

If you lift and shift, which is very painful, and then just stay lifted and shifted, it's much less green than being in your own data center. Because in your own data center, you are limited by things like available number of plugs. Whereas one of the things that Azure or AWS, or whatever, takes away from you is the worry about the available number of plugs. So you can just put in machine after machine, machine and you over provision like crazy. So the cloud is massively dangerous. If you just lift and shift into it and then you don't update all your systems to take advantage of all the auto scaling in the cloud and the services that can make sure you don't over provision, you will over provision. But if you do go all the way to the end and you can use multi-tenancy, multi-tenancy is marvelous because yeah, you've got people who are doing things at one time a day using the same hardware as tenants who are doing things at different times of the day. multi tenancy is like magic.

Sara Bergman (23:47)

Yeah, no, I agree. I think it's also about, because like we said earlier in this episode, all cloud providers have the money and the people to spend on code efficiency and like being smart with how things are placed onto the physical machines. Like that whole layer is managed, of course, and all cloud providers are interested in that being as efficient as possible. but the users have to help them, by enabling auto scaling or by running in a right size container or like not provisioning a whole physical machine, cause that makes it more difficult. Of course, if that is what suits your scenario the best, that's of course what you should do. But if you only do it because that's the easy choice, then maybe spend some time thinking about it before you decide.

Anne Currie (24:36)

Yeah, dedicated servers are evil.

Sara Bergman (24:39)

Yeah, yeah. And I think one of the, there was another QCon conference talk la few years ago now from Duolingo. They had a Super Bowl commercial where the TV screen said something like, do your training, no bots. And then they also wanted to send a notification to a large set of users in North America who were supposedly, hopefully watching Super Bowl. And to coordinate that massive, cause that was a one time thing. Like they were just going to send it out for the Super Bowl, needed a massive capacity to do that. And then they could scale down. They had a great talk. It was so interesting to hear how they worked together with, I think AWS were their provider and like how they talked about it. Like how will we make sure that we can scale up and scale down quickly? Cause we don't want to spend a gazillion dollars on having all this capacity for longer than we need. Cause we just need it for a few minutes basically. And that was super cool. And that's one of the benefits of the cloud, I guess. But yeah, now we talked a lot about the cloud. We were not supposed to talk about that.

Anne Currie (25:53)

Yeah, so I feel every time we talk about the cloud, I need to put in a warning, which is, and this is very much a warning from my past when I used to be a head of IT. The senior folk at a company often just don't understand. They're sold "cloud good" and you as an a head of IT or a CTO, or whatever, will also say "cloud good", because you're thinking about the long term - cloud native when it'll all be fantastic and massively efficient. But because you do it in two stages, the first one is unbelievably painful and just gets you lift and shift. And then the second one gets you cloud native.But after stage 1, all the senior folk like the CEOs go, well, we're in the cloud now where are all these benefits that you said they'd be? You know, we'll not listened to you again. So yeah, it's very difficult to make sure that you frame it as a two step process with your senior folk and get their buy in early because you're going to blow all your credit doing a lift and shift and it's going to be hard to persuade them that no, that was only halfway because..

Sara Bergman (26:50)

Yeah, you said cloud good.

And I think that's also something, because I did talk about running on ARM chips and potentially having your software be able to run anything at the conference. And I realized that's maybe step three in this journey, because I think that also unlocks really cool things where you can also be flexible for future demands. Because if there's one thing we know is that chips are forever evolving.

These companies are doing really cool stuff and you don't know what's going to be best. And you also don't know what's going to be cheaper. You don't know what you might need. You might be fine running on an older, slightly less efficient chip for one workload. But for another, you want to be really agile and constantly be jumping. so doing that can give you a green wind. Also, if you are on-prem you might want to change from X64 to ARM64 long-term, but changing everything you have in your data center at once is not very green because hardware is incredibly expensive from a carbon and monetary perspective. But if you have code that can run on anything, whenever you pop one out, you pop a new ARM one in, for example, and you gradually transition. I think that's one of the cool benefits.

Anne Currie (28:32)

You make a very good point there that things change. One of the greenest things you can do is take new tools and services and hardware and things - be able to take the better stuff that comes along and that is an operational skill. ⁓ You know, you need to be able to to move to new things.

Sara Bergman (28:51)

Yes.

Anne Currie (28:58)

Arm is a really interesting example of the evolution of things. I mean, it seemed a few years ago that Intel had just got it nailed and no one would ever displace them.

Sara Bergman (29:10)

Yeah, no, think it's interesting. You did a podcast with the chief evangelist at Ampere for Environmental Variables [podcast] and that was great. I've listened to it multiple times because I think it's so good. And he's saying some really cool things. Even though I work with CPU, you kind of still think of them as a black box. Unless you're building the CPU, they're still like... or slightly transparent, like foggy box, I don't know. And there are things they're doing with like cache placement . That they are building their CPUs intended that the sweet spot should be 80 to 90 % utilization. That to me is like so cool. And that can have massive savings in power, massive savings, energy efficiency, because arm chips are designed for low energy environments. They're not new because they've been used in low energy or energy constrained systems for ages. We just haven't really seen all systems as potentially energy constrained, which we might start to do in the world we're going into. I don't know. think it's another thing where think both you and I think in the future, we will not be energy constrained. We believe in the energy transition and clean energy for all. However, that's not going to be next year or two years from now. So we have to be mindful now and then in future we'll have clean power for all. ⁓

Anne Currie (30:41)

Yeah.

Even with batteries getting better and better, solar generation and wind generation are still getting better faster than batteries are getting better. And the sun is not getting better. So fundamentally, the days are getting longer in a sense, but we're never we're not gonna have...

Actually you occasionally do have 24 hours of sun.

Sara Bergman (31:07)

Yes, in a very small part of the world and the other half of the year it's very dark. So that's when we need the most electricity.

Anne Currie (31:16)

But for most of the world, it's kind of like 12 hours of sunshine, 12 hours of dark. And if you can shift work in time and delay it or bring it forward, then you will have an incredible amount of power. But I do think the flexibility comes into that massively.

Sara Bergman (31:37)

Yeah, no for sure.

Anne Currie (31:38)

Yeah, yeah, asynchronicity. 

Sara Bergman (31:41)

Exactly. That should be another episode just talking about the energy transition because that's another fascinating concept.

Anne Currie (31:47)

we, do. Yeah, so you and I are going to have to line up loads and loads of episodes now before your baby arrives. 

Sara Bergman (31:55)

Yes, because again, we are asynchronous and unreliable.

No, but another thing I think is interesting also with code efficiency and operational efficiency is also different languages are differently skilled at this because some are built for this scenario. Like we talked about Rust and I've never really used Rust, but that's becoming increasingly popular. And as I've understood it as well, they're really good at this.

Because for other languages, you might have to compile twice if you have ahead of time compilation. And that might be fine if your CI pipeline is not run a gazillion times a day. If your CI pipeline is run a gazillion times a day and you start producing two artifacts, like that's another carbon cost and general costs you have to take into consideration. So there's always these trade-offs.

Anne Currie (32:48)

Yeah, Rust is a really efficient language and it has all that memory safety as well. I've never written anything in Rust. used to write a load in C because I'm old, but I've never written anything in Rust. I'm not inclined to learn a new language now, especially one that's tricky as Rust even though I find it really interesting, because I think that... Again, I may say this you couldn't possibly comment. AI is changing the balance of of wins in writing new code and AI is so good combined with Rust because the Rust compiler is so opinionated. it's checking what you do. checking what the agent does. Which is useful. So there's a lot of code being written in Rust by AI. But yeah, I think I'm too lazy now to learn a new language when I want to find out what's going on with AI first.

Sara Bergman (33:40)

Yes.

Anne Currie (34:03)

But at same time, I'm very aware that there's that danger that everybody's stepping back and going, I don't really know what to learn next, because I don't know what tools are coming. I don't know what the situation is going to be. We do keep moving forward.

Sara Bergman (34:17)

Yeah, I know that's true. I think there's something also for just the joy of learning new things. we have hackathon weeks at work and I do always enjoy it because I get to rediscover that joy of learning something new, which I, as a geek in case that wasn't obvious to everyone, really enjoy. That's also why I enjoy conferences and like the hallway tracks of conferences because you always learn something you might have not expected to learn.

Anne Currie (34:42)

Yeah. well, that's a good segue that we should actually talk about. So we won't talk much longer because we've already covered so much. Everybody's heads will be on the verge of explosion. But there is something that I think I want to have as a constant theme in the podcast, which is to undermine my own my own sell which is podcasts are great. I love podcasting. And I like a video conference. This is 50 % as good as seeing you in person, which is so much better than not having any interaction with you at all. But seeing people in person is a step up again. 

Almost everybody that is a guest on this podcast I've met in real life at a conference, probably the first time. Wandering around watching talks is a very good way of meeting people and discussing things and achieving your network and especially if you talk.

Sara Bergman (35:48)

For sure. No, I agree. Also, the thing about podcasts is not everyone can and wants to be on a podcast either, but you might feel very comfortable at the conference.

Anne Currie (35:58)

Well, yeah, no one feels comfortable talking in a conference.

Sara Bergman (36:01)

No, mean in a hallway, in a hallway track, you don't have to be on stage, but you can just be a conference attendee.

Anne Currie (36:08)

Yeah, that's true. You can just be a conference attendee and stand around and listen, and even just standing around listening, saying the occasional thing you will meet people. If you're interested, people spot that, and everybody's a bit nervous at a conference, which is actually good. Because when people have a little bit of fear and nervousness, it means that if you smile... The best tool you can take to a conference is a smile you know. Look like you will be happy to talk to people and people will come and talk to you.

Sara Bergman (36:42)

And not move about in a herd. Because that's very difficult for anyone. Also, I'm Scandinavian, an introvert, I would never approach a group of people because that's terrifying to me. One or two people, like I can talk to them.

Anne Currie (36:59)

Yeah, that's true. It is really hard to approach a group. Much, much easier to approach a person. 

If you want to meet the super big cheeses at conferences, it's good to be speaking because everybody who's on stage is super nervous. And therefore they'll bond with you over your nervousness about being on stage. You don't have to leap there in one go I would I strongly recommend that that people think about speaking at meetups, even if it's just five minutes. Yes, even if it's just doing something like hosting a panel, so you don't have to say anything yourself, you just get used to being on stage. I remember when we first met, virtually met on a call, and you said you were interested in getting into conference talking. and you felt quite comfortable being on stage because you used to do a lot of dance, which is on stage. So you're already confident being on stage. Then it's just a matter of being confident saying your own material, which is a good pathway.

Sara Bergman (37:52)

Yeah, so put your kids in dance class, folks. They'll become amazing conference speakers. Just kidding. No, I think that's a good path. And I also did, I did the meetups first, like you said, because that's, it's often also a smaller, more intimate crowd, which can be less daunting, but also you can get really good feedback. Also before I do a conference, I have like set of people I work with. I always say like, I'm going to do a friends and family edition of this talk. You can come watch it if you want to. No pressure. And then I know I will get feedback. My talk always looks different after that friends and family edition with the people I trust. Cause I know they will give me honest feedback. I want that and I need that. And that, like, wasn't comfortable the first few times, but now it's something I value so much.

Anne Currie (38:54)

So now I'm going to do something which I said I I shouldn't do. I'm going to refer to something that's happening in real life quite soon before we recorded this session, which was the Oscars yesterday. So this may not go out for weeks or months, but fundamentally in our timeline, it was the Oscars yesterday. Way back before I started talking at conferences, I had a - it's like my hobby is painting and I had a job teaching a Hollywood Oscar winning actress how to paint like a naive painter for a bio picture. Yeah, Have I told you this story? 

Sara Bergman (39:31)

No, no, this is awesome.

Anne Currie (39:42)

So yeah, it was Sally Hawkins, who had the Oscar for best supporting actress for a Woody Allen film. And then the year that we worked together, she was up for best actress, she didn't win. But anyway, although she's brilliant in that film that she was up for. But, while we were working together, and she was learning how to forge this artist, she was presenting an award at the BAFTAs...

Sara Bergman (40:01)

Yeah, the British film awards, right. Yeah.

Anne Currie (40:07)

And so I watched her and I watched the whole BAFTAs, which I don't normally do. And afterwards I talked to her about it, and said, was it my imagination? Or was everybody on stage - And these were big names, Tom Cruise was the big host. - was everybody really, really nervous? And I don't think I would have noticed that except that I was seeing her every day, and I knew what she was like normally. And then I saw her on stage, and I thought, my God, you're incredibly nervous. And then I realized, everyone else is incredibly nervous. They're all megastars. They're on stage and they're incredibly nervous. She said, yeah, we're all massively nervous when we're on stage doing something like that. But you're la stage actress, you're an award-winning stage actress. Why are you nervous being on stage? She said, because when you're on stage as yourself, everybody is nervous. Everybody is unbelievably nervous. Tom Cruise is nervous. Oscar winning actresses are nervous. Everybody is phenomenally nervous. And when she said that, it made me realize that I had been kind of thinking, well, you know, I shouldn't be on stage because I'm obviously not supposed to be on stage because the thought fills me with fear and nerves. Therefore, you know, I'm not naturally somebody to be on stage. But then after she said that it made me realize everybody is massively nervous to be on stage. It's not a reason not to do it.

Sara Bergman (41:33)

No. Like if you do it and you feel absolutely horrible afterwards as well, you know, maybe this is not a great option for you. But most people feel amazing afterwards because you're like, oh my God, I did it. And it went well. And then people come up to talk to you afterwards. You're like, oh my God, you listened to the words that came out of my mouth. And like that feels really good. And you get to meet other people. You get to see other parts of the world maybe or even other parts of your own country. And now it's just good vibes.

Anne Currie (42:00)

Yeah, and start easy as you say, friends and family.

Sara Bergman (42:05)

And local conferences. You don't have to travel across the globe for your first one. there are many, many, I'm sure, in whatever country you are. I have been speaking at quite a few in Norway and Sweden and Denmark, which is very close to where I am as well. Which is also nice because then I get to meet people who are closer to me otherwise and then I get to be on Swedish podcasts and whatnot.

Anne Currie (42:33)

I'm very aware that we could talk forever, but we shouldn't because we'll burn out all our listeners. And we have plenty of things to talk about the next time we do a podcast, which we need to buffer up. So thank you all very much for listening. There will be show notes associated with this and I will try to put in all the links to the things we talked about today, like the interview I did with Sean Varley for environment variables. Also I'm very keen on people giving me feedback and asking questions that we could talk about, you and I, or maybe other guests as well.

Do reach out to me on LinkedIn or on Bluesky with your questions. I'll be extremely happy to see if we can answer some of those questions, talk about some of those things in future podcasts. So thank you very much for listening, everybody. And thank you very much indeed, Sara for being a guest.

Sara Bergman (43:47)

Thank you so much for having me. It was a pleasure.

Anne Currie (43:50)

And I hopefully will I'll see you again, Sara, soon another episode and I'll see everybody again hopefully soon on a future episode of asynchronous and unreliable podcast. Thank you very much.