S15E08 The State of the Power Grid with Mike Ratliff
===

(00:32) Charles: Hi everyone, I'm Charles Suggs, software developer at SmartLogic.

(00:36) Emma Whamond: And I'm Emma Whamond, also a software developer at SmartLogic, and we're your hosts for season 15, episode 8. We're joined by Mike Ratliff, co-founder and CTO of GridVar. Mike has spent over 30 years building software, leading engineering organizations, and has focused much of his career on utilities, distributed energy systems, and electricity. In a season focused on how technology is adapting to a changing landscape, we're looking at one of the largest and most complex systems. the power grid.

(01:10) Charles: Mike, welcome to the podcast.

(01:12) Mike Ratliff: Hey, thanks for coming on. It's great to be here.

(01:14) Charles: Yeah, nice to have you. ⁓ you you've had a pretty fascinating career so far. Maybe you can give us some of the the cliff notes of that that journey and how you got into the energy and utility space and what you're doing today.

(01:27) Mike Ratliff: Yeah, it has been kind of a winding path. I think the you know the front half of my career was all about ⁓ kind of exploration around the software domain a lot of time working in different product and consulting companies, kind of getting a feel for what was out there. then when I got an opportunity to work in the energy space, it really wasn't my plan initially to get into that. as soon as I started working ⁓ on some problems in that space and kind of a word small area. I kind of realized that it was one of those industries that was ripe for a lot of change and it's probably like what the banking through and telecom industry went through in years were just gonna be a lot of things that were gonna need to be thought about and worked so it just became a little bit of a passion. And then as I wrapped up each project or each company, just kinda naturally started to roll into other opportunities there because new coming about on the grid all the time. And, you know, it seemed like a never ending like ⁓ faucet of just opportunities to to look at these kind world changing problems. You know, we as you know, that we all use power as a species in in really different ways are changing pretty dramatically over the last decade or two.

(02:43) Charles: fits with of the the state of the stack and how everything is a lot is changing. ⁓ yeah. For I suppose some would say good and some would say bad or both.

(02:45) Mike Ratliff: Hmm. Yeah. It is. Yeah, I don't think anything's ever, you know, a hundred percent one way or the other on that front, so you'd kinda have to tease it apart there and take the good with the bad.

(03:01) Charles: Definitely, yeah. and you've worked through several waves of change in our industry from of the early internet era to cloud computing and and AI now. What what shifts have felt most significant to you?

(03:16) Mike Ratliff: Yeah, it's a it's a really easy question to go right to like what's happening today because everything seems so dramatic and and all the changes that we're going through. ⁓ you know, I do think that there's there's been some interesting parallels in all of those phases. And you know, it's probably become a little bit more in focus right I I think one of the things that that has been interesting to me just to watch over those changes is that one of the consistencies is taste and insight about what's going on and how to apply that to what's really in our industry. And that really hasn't changed a it maybe has had a little bit more of a magnifying glass put on it and what we can all do today the breadth that we can we can apply that taste. you know, I think that has been one of the interesting kind of consistent pieces that each one of those phases has gone through is You've seen a lot of people spend a lot of time and effort building things that ended up not really mattering in the end. ⁓ and some of that you could probably look back with the benefit of hindsight and say, if we'd thought a little bit harder about this and like really applied some thinkings about what's going on in other industries and what's happened, you may have been able to avoid some of those things. And I think that's just such a timely lesson for what we're experiencing now, because it's so easy to build things that really it's coming down to having that insight to say, should this be if it should, should it be built in this particular way? And so people that are gonna f that figure that out, I think are gonna have a lot of fun in this next round of changes.

(04:51) Charles: think some of that is about like deferring engineering decisions until know it's something you need to be focused on, or is it also about really kind of thinking through a lot of those possibilities to know what makes sense to defer or to not build yet?

(05:06) Mike Ratliff: Yeah, that's a really interesting way of putting it. And I would I guess the the knee-jerk reaction that that I have to that is maybe we don't have to defer quite as many things as maybe we should have a few years what we can actually explore of very, very high is fairly astounding And so, you know, what I've found in this startup that I've been working on is don't really to defer a lot of hard engineering problems. We kind of tackle them head on. And even if we have to end up throwing away something, it's not a very large investment in terms of time and order to do that. and you just have to have the willingness to be able to let go and be like, yeah, that was a I was a great learning experience, but it's let's just nuke it all.

(05:50) Emma Whamond: ease of building one thing. What else would you how would you compare the industry today to when you started in the nineties? Like what feels fundamentally different?

(06:02) Mike Ratliff: Yeah, I think the the breadth at which we can have influence is just completely different. ⁓ you know, was it was in vogue back in, you know, the nineties and even the two thousands to have, you know, fairly specific titles, you know, database engineer, end engineer and whatnot. just seems kind of prosaic because all the people that that I work with that are really having the most fun and seem to be having most kind of the generalists. And that's ⁓ I think that's been an an interesting thing for me to kind of come to terms with because I think for most of my career was a generalist. know, I try I experimented with a lot of different things and I was never the best at any any one I was always kind of looking back, going, God, am I making the right decisions? If I should I have gone really deep into one particular thing and become that And then today happened with all of the, you know, the tooling that we have at our fingertips. And I'm so grateful that was kind of exploratory and kind of creative with my choices it's that ability to kind of connect these weird, seemingly unrelated dots that I think is a lot of fun today and what and where people are becoming really, really being able to do to pull those things in. And and those are the folks you really want to work with today are the ones that can make these connections you're like, wow, where did you come up with it's because they did some random thing ten years ago in their life that they then it all comes together. And ⁓ that's been super cool to watch happen and watch other folks kind of have that aha moment that something they thought they thought was a idea now becomes core to something that that made business decision actually work.

(07:55) Emma Whamond: I love that answer. It certainly is the age of the generalist and the people that are willing to be generalists,

(07:59) Mike Ratliff: Yeah. Yeah, absolutely.

(08:04) Emma Whamond: are there any aspects of the industry today versus the 90s that feel surprisingly unchanged?

(08:12) Mike Ratliff: Mm-hmm. taste still still matters. I mean I I kind of keep coming back to that, but it's you know, I I think we have a very significant shortage of people with really good taste and insight. And ⁓ and that's no different today as it was before. I think it just it maybe even get mag gets magnified a little bit more today just because people can deliver things much more quickly. So very much in your face if somebody's built something that that isn't quite there. ⁓ but yeah, I would like to think that that has been kind of the the unifying human thread throughout all this we're just getting a little bit better at all the that's my hope anyway.

(08:55) Charles: When you and when you build the wrong thing with with real world infrastructure, it's even more kind of in your face and maybe lasts longer too, ⁓ given the the physical nature of some of that.

(09:00) Mike Ratliff: Mm. Yeah, it's been a c several of the projects that I've worked on have had aspects where they go out and they touch physical devices, whether that's the the entire grid itself or it's just physical ⁓ devices within buildings and things like And that's a very fascinating way to to build or aspect of building tools because you know the the hardware world and the software world don't exactly line up in the way that you go about building things. And I'm not a hardware engineer by trade, so I don't want my experience is is spectacular here. But I've had to learn how to lead hardware engineering teams. And as a result, I've learned a little bit. And a result, of being with people on that side of the And you know that bringing together of those two different cadences and two different levels of kind of how rigor gets achieved been really fascinating to watch because you you get a lot of chances to make things right in software and so you end up be getting this little bit of a cavalier attitude sometimes with hardware you get a couple of chances to get it right. And once you've got ten thousand, a hundred thousand of these devices out in the world, if something's really wrong, it bankrupts your company. and that is it just it it requires a different level of thought and careful rigor up we've kind of eschewed in the software world and we kinda we use a lot of other techniques to kind of get by on that when you try to bring those two worlds together, it is an interesting cultural mishmash, to say the least.

(10:54) Charles: digging more into kind of the utility and energy space where you've spent a bit of your career and particularly more recent, what what makes building software for utilities and these energy systems different from building software for a SaaS application or something that, you know, you just deploy to a server and if it doesn't work, you can deploy again and again?

(11:15) Mike Ratliff: Yeah. I think the you know, utilities have they're by nature one of the more risk adverse cultures that you will interact with out And you know, that makes it very hard to ⁓ engage with them, especially as a small startup and things like that, because a you know, they have to bet on these very long term plans. And so the the one thing that I've found that has been helpful in kind of getting software ⁓ to get that that kind of flywheel nature of getting engaged with is you you can never really think about trying to sell software to a utility is actually gonna alter something on the grid right up front. they're gonna wanna go through this phase where they understand like you kind of go through this like data collection. analysis, prediction, and then control level of evolution. if you can find a company and a business problem that has some beats on each one of those that has value that you can that you can deliver, you can kind of escalate up to that level of actual control. and that's really hard to do and it's a it's a long journey and process. But know, I think is the you know the current opportunity that I'm working on. You know, we're really going through that today of like how do we add value without running into some of those really real valid roadblocks that utilities are going to put keep the grid stable. Yeah, they need that and they need to be skeptical.

(12:50) Charles: So what kind of work are you doing with GridVar that you think would kinda enable you to work your way into that that flow of change that utilities work with?

(12:59) Mike Ratliff: Yeah. So what we're trying to do, and everybody's probably heard this, like there's this problem today where connecting large loads to the grid is a huge problem. And it's a problem for a couple different One, there just isn't the power available. ⁓ and so if you need a you know a gigawatt for a data a gigawatt of real power is a challenge. But Kind of the answer to that problem is really relatively easy. It's like, is there a gigawatt or not? ⁓ that's not the real problem with interconnection, although that's what you hear about, like the lack of power. Where these developers are really struggling a lot of times, is they may find the power or they may procure the power themselves, and then they go to to actually interconnect and make it deep in that cycle. They've committed a ton of capital. And then they find out that the power quality problems that they're going to induce on the grid have kicked them out of the queue. they have voltage excursions, they have harmonic problems, they have all these other things. so that's the problem that we're trying to really attack. ⁓ and basically build software that can simulate all of those problems from a power quality perspective that you would normally have to pay, you know. Hundreds of thousands of dollars and wait months to get those studies done. built software that allows a non-technical user to get those answers in minutes. And so they can sit there and experiment with a lot of different ⁓ approaches and then, you know, find a good connection strategy so that they know exactly what they're getting so from that perspective, that is that kind of collection and analysis and prediction step where we're not actually doing the control side of the grid doing the on ramping side and so your risk tolerance is a is in a di completely different place.

(14:51) Charles: In in Illinois, we've had a a lot of legislation to promote expansion of solar and such ⁓ among other things. But one of the hurdles we keep hearing about is

(14:57) Mike Ratliff: Mm. Yeah.

(15:05) Charles: grenader connection approved by either the utility or the grid operators. ⁓ and it sounds like what you're working on might alleviate some of that if if it's in especially due to like power quality problems.

(15:19) Mike Ratliff: Yeah, and I I think, you know, it is an area that a lot of people are looking at and as I've know, I've been working on this particular company for about nine months and we're not the only ones. Like there's a bunch of people looking at this there's a ton of money you know, being locked up by this. And gives me a lot of hope that a lot of smart people are looking at these problems and trying to find some good solutions it's broken in such ways that I I it it it boggles the mind. at the at the scale at of of backlog that is out there. ⁓ it it's astounding.

(15:55) Emma Whamond: So I I heard that you said utilities are slow to adopt a new technology. It's not entirely irrational, you know, when a when a SaaS company has an outage, people complain on X and ⁓ when a utility has an outage, you know, people die in extreme weather or traffic lights fail, hospitals have to run on backup power.

(16:00) Mike Ratliff: Yeah. It's not Yeah.

(16:16) Emma Whamond: In the technology space, there's a parallel there between moving software towards distributed systems and then the grids move towards a distributed energy. So when we talk about distributed energy, could you give us a rundown of what you mean by that?

(16:34) Mike Ratliff: Yeah, I mean I think the best way to that I understand it is to think of it that for decades the grid has operated really well by this centralized control mechanism and where generation we can say generation chases a load. And what that means is that if everybody consuming power starts consuming more power, all the generation ramps up to service staff. this model has has treated us very, very well for many, many decades. the reality is it's it's starting to fail and utilities are seeing the that and it's failing because the the level of dynastic of the loads is so it's hard to chase those loads fast enough. it's also it's also a challenge that the cost to actually deliver power at any moment at any level. is astronomical. I mean so we have this huge amount of waste of overbuilt capacity so that we can ensure the grid And you know this is all shifting. And so when we think about distributed energy, lot of things I think about are how do we invert that control? And that inversion comes on a couple different axes. One of them is that ⁓ load should follow generation instead of the other way around. And that really means a lot of times having ⁓ cost signals or price signals that ⁓ change people's behavior so that they consume power a more responsible way. And it also means that we are generating and consuming and storing power out at the edge of the grid instead of at the centralized part of that means that the the classic model of centralized control just really can't play anymore. you know, if you go into a utilities, you know, operating center, you know, a lot of times you'll see like I mean there's an enormous amount of information on there, but there's still humans in the loop making decisions. And that's great when things change by minutes and hours, but it it completely fails when things change in subsecond cycles. And and that's the reality that we're going into the machines are gonna have to control large aspects of this, and that's terrifying to ⁓ an organization that has traditionally not turned over that level of automation.

(18:56) Charles: What does that look like? Is that like devices at the end users that are doing some kind of demand side response that maybe moderate some of their load at peak times for the grid or when it's advantageous to dial something back or maybe to ramp something up because there's too much on the grid and they need to vent off some some power. ⁓ go ahead.

(19:15) Mike Ratliff: Yeah yeah. No, yeah, you actually I I like the fact that you went both directions with that because a lot of times people just think about ramping down ⁓ consumption. But we're in a world where sometimes we have an overabundance of power, especially where there's a heavy influence of wind and solar. know, there's this thing called a duct curve that we talk about in the power industry where it's this w the way the power ⁓ signal looks when you when you have a lot of solar co you know in the afternoon that that drops off as soon as know the evening starts to come on and all of the air conditioners and people come home start to come in that mismatch has been a real challenge. so yes it is really about having devices that can respond to the realities of what's going on in real time and not have a human have to deal with that. if you do that in in each individual place and you have some Some relatively reasonable guidelines. it's just a it's a very complicated system, but when you break it down, it's very simple at the individual nodes. There's like, hey, behave well here. know, if everybody behaves within those realms, can start to self-heal a little bit more. But that, you know, that world is very slow to come about. You know, people don't want to give up their their their control, utilities don't want to give up their control. And so the we're we're kind of stuck in this this very rigid case where people see the solutions that need to be out there but don't always follow through with making them real.

(20:49) Charles: The elixirist in me picks up on self-healing here as talk about the are there lessons that s ⁓ software developers can take from, say, the world of distributed energy and apply that to distributed computing?

(21:04) Mike Ratliff: You know, it's weird. I would almost invert that and say maybe the the world of software developers, especially even the Elixir world, has got a lot of this figured out and that that needs to flow into energy side of things because we have been holding very tightly onto this centralized control metaphor for for so long in there. And so I think it actually flows that direction right now. I'm not sure. are a lot of lessons that we can pull from the ⁓ the grid itself right now in terms of distributed and things like that. I think that's the weakness in the in the grid today.

(21:44) Emma Whamond: So with rooftop, solar, home batteries, EV chargers, microgrids, virtual power plants, all the are those other aspects of the distributed energy system that go into all of this instead of just being taken, also giving energy, kind of goes back to the distributed system of software.

(21:52) Mike Ratliff: Yes. Yeah, I mean ⁓ microgrids are a great I think a great tool and you're probably gonna see more and more of those come about because know they you can kind of abstract a a lot of nuanced behavior behind this point of interconnection and if you get them in agreement with the utility to say like we're gonna make this point of interconnection within these parameters. and that that just creates a a simplicity and moment in a large complicated network. And you know, I think that is something that you're seeing a lot. lot of the the hyperscale data centers and stuff that we're starting to model for interconnection, of the big problems is like the the level of swing that they create on the grid in terms of voltage and harmonics and things like that. if and so if if you can come to the utility and say, what's going on behind our meter, our point of interconnection is all over the the board. But we're gonna put in the gear to make it look really the point of interconnection. utility's like, ⁓ thank you. I have one less thing to worry about. that I think that that like tool right there to think about these islands and these these moments of stability that everybody's driving towards, gonna be what starts to unclog some of this backlog of the interconnection. when I talk to power systems engineers, That's what they're most terrified of is the like plugging in these hyperscale data centers and they just don't know. don't have the math to really describe what's gonna happen in a subcycle moment when, you know, those GPUs dump all their memory they go from a gigawatt to a hundred megawatts of power in no time at all and what what occurs there. y you see it even in smaller cases, like, you know, if you plug your EV into your house and things are a little like You can t actually hear and see the harmonics sometimes in the lights and things like that. real. I mean we see that and it's not even localized. We see data centers providing those harmon hitting those harmonics that are being measured hundreds of miles away. it's it's amazing.

(24:15) Charles: We're all connected.

(24:18) Mike Ratliff: We are and it's ⁓ I'm honestly astounded every time I flip a light switch and the power comes on. Like that's that's how terrifyingly fragile I think things are getting.

(24:32) Charles: That brings an interesting question about cybersecurity on the this is something that's been in the news, of course, for for a while now. Does distribution make that situation better? Does it make it worse or more complicated? And what role does software have to play in in that?

(24:50) Mike Ratliff: Yeah, I mean, so I'm not a cybersecurity expert. but I think that anytime you talk about these pendulum swings from to distributed, it's it's not about it being better or worse. It's about trading one set of problems for another set of problems. and you may like some of the problems on one side of the pendulum swing better. and I think what's nice about the the move that we're going through now is that Any individual problem is becoming less and less impactful. if you do have a microgrid, for example, that has, you know, one point of interconnection and you've managed that, even if something you know goes on you get one individual thing gets hacked inside of there, odds are the control systems for the other devices and stuff will be able to make up for that in some way. And so I think that's better. But man, the the attack surface has gotten really, really large. ⁓ and that's I think what has gotten most people frightened on that front. so, you know, I just think in general that attack surface problem is really interesting right now because you see what's going on with the AI models being able to find vulnerabilities. I think we're going through this like six to eighteen month time period where things are just terrifying. but then we will probably harden a lot of our just tragic mistakes that we've made over the last few decades and maybe get off out the back end of something where our systems are more resilient at that point. But the next eighteen months is gonna be an interesting ride.

(26:22) Emma Whamond: So I'd like to look a little bit further in the future with climate events, placing an increasing stress on the infrastructure system. How are utilities adapting to a world that might be less predictable than it was twenty or thirty years ago?

(26:38) Mike Ratliff: So I even though I've been in the in this space for sixteen odd years, you know, I've never actually worked at a utility. I've sold to utilities and I've you know I've got friends that that work there. so, you know, I I think my answer is gonna be a little bit of an arm's length distance from that. I do think that the paranoia and the cautiousness is both working for and against the utilities on that front. ⁓ I like the fact that there's a lot of thought being being placed into like how does the system actually respond all of these different contingencies that they utilities do a great job of simulating a massive number of contingencies. so that's a that's something you can kind of take a little bit of comfort in. But the ability to unlike to simulate every possible failure is just not real. And I I think if we could put one thing in place that would maybe start to help that, and we're starting to see it, is just the generation and load link that has always had to be completely in sync. And you know, that's where you know large scale storage is really going to come to the rescue of being able to kind of create a buffer in how all this stuff operates. It's not there yet from a cost perspective. as you start to see more and more of that come come online, I think you're going to have ability to absorb more and more of these changes in a reasonable way. that is probably one of the we need that shock absorber in the power grid. And if if I were to bet on one thing, it would be wide scale storage deployment that would get us to that. ⁓ a little bit more of a safety margin there.

(28:20) Charles: I've read that when industrial scale solar facilities are having trouble with grid interconnection or they have some but they are not able to get another facility online. Well, if they add battery storage at an existing facility and expand add more panels there, now they've flattened their load curve where it's no longer peaking when it's sunny and dropping at night, but they can with the battery storage just smooth it out.

(28:45) Mike Ratliff: That's exactly it. nailed it. the storage combining with renewables is really the key. because it has been a as much as I lean very left in my politics and thinking, a realist about where we are today and the baseload and the peaking load generation, you know, there's a lot of natural gas that we burn it's necessary in order to keep things balanced out we can get exactly what you said to be ubiquitous. And at that point we can really start to see become a much more like reasonable player in terms of you know long-term, you know, 24 by 7 energy storage. It's it's really hard. I mean we have we built software here at GridVar that can simulate building microgrids. And I've gone through and built thousands and thousands of different microgrids with this and what's disheartening is that it's really hard to build something that can operate twenty four by seven without burning natural gas.

(29:47) Emma Whamond: So it's our understanding that as CTO you played a role in bringing Elixir to the company. could you give us a rundown of some characteristics of Elixir that made it a good fit for energy and utility applications? I talked about distributed energy, but anything else you could speak to?

(30:07) Mike Ratliff: Yeah, I brought Elixir into now like five or six different companies. it's kind of been my go-to tool set for ⁓ a lot of these a lot of these projects. And know, I I think for me a lot of it came down to the peace of mind of being able to go to sleep and and knowing that there's a a layer of protection like if you do it you know, you even if you have defects that things are not gonna completely go sideways and that you know you can get a lot more self-recovery. think that's that's been really hugely beneficial and I think the other thing is that just the the nature of how it seems that the the beam is a really really good job of maybe not being really never falling over even in the midst of like stress or more data flowing in and things like that. And that has been really important to me because you I build systems that run twenty four by seven. You don't like get a break. so having something that has that ability to have some back pressure and some recovery around processing when you're just getting a fire hose of telemetry coming in the time, been really ⁓ a great piece. then, you know, it it it's hard to it's hard to ignore the just the productivity that you get around ⁓ the ecosystem as well. I I try to I try to keep up with what's going on in the tech world and try to keep an open mind with other frameworks and things like that. And I gotta be honest, like I I still I still haven't found anything that solves these problems as well. So open minded about it, but I haven't seen anything out there.

(31:45) Emma Whamond: hear that 'cause we agree. Yeah. Lightweight processes, supervision trees, fault tolerance. yeah.

(31:46) Mike Ratliff: Mm-hmm. Yeah. Yeah. Yeah. Yeah, there's a lot of places, there's a lot of ⁓ stacks that are seem to be trying to pull some of the superficial benefits of the beam. and it's it's really kind of disheartening to read about that and you're like, Well you took the names, but you don't really have the real piece behind it, and I don't know if everybody reading that truly appreciates how different some of the implementations of some of those features are they try to pull them into another runtime.

(32:23) Emma Whamond: We interviewed someone last week that was teaching others how to bring Elixir into new companies. this is the perfect segue. Were there any challenges unique to applying it to an energy or utility space?

(32:39) Mike Ratliff: Yeah, I mean, it wasn't too bad. ⁓ you know, I'm I'm fortunate in that, you know, I've been able for the long time I've done CTO level roles and so you get a lot of latitude in being able to make these decisions and so you don't have to manage up quite as much bringing the technology in. I will say, like, when a utility goes through due diligence on your platform, know, you hope that it's primarily about penetration tests and things like that and deployment architecture and and all that and that is a a big part of it. but there's always a question about, you know, like what's this made in? And for me, it it really you do, I'm upfront about it. It's like, hey, we built this in Elixir, here's why. I just immediately go into, we know that you need this to run 24 by 7. You need it to have this level of resilience. Here are the things that this that this framework gives you. here and there's some good references out there now that we you can point people to that are that are at least semi-technical to be able to understand. honestly, it it really hasn't been a a sticking you know, I think the biggest sticking points that I've had in adopting Elixir is when I've come into companies before that were maybe kind of down the road and they brought me in to like maybe fix things. And you know, I've I've begrudgingly done a couple of pretty significant plat replatforming to Elixir. say begrudgingly because that's always a terrifying and annoying thing to have to do. sometimes it's necessary. And you know those are the those are the hard ones. not so much about a hard discussion about why Elixir. It's more of a hard discussion about why are we going to spend all that money and can you prove to me that this is really going to be a better way of doing things. And and that's that's the hard equation, but that's not an Elixir specific discussion.

(34:27) Charles: What ⁓ I guess it's n what tends to kind of Get people to buy in who are skeptical about doing a replatforming and

(34:38) Mike Ratliff: Yeah. ⁓ I mean it it's all it's all about money. ⁓ and it all has to come down to that. And you have

(34:43) Charles: Mm-hmm.

(34:46) Mike Ratliff: have to be able to paint a picture of from every angle. Okay. Here here are what the customers are saying, here are the frustrations they're experiencing, what sales is unable to do, where marketing has to paint a a really interesting story. you know, here's where you know and you d you just go a across that and you and A company knows where those pain points are, when you pull them all together and you create a cohesive narrative around why all of the woes that a company is are is experiencing all pointing back to s a know, a platform or a tool or whatever that is that is causing this friction, you can stand a chance. And you know, I'm a big believer in the in storytelling and as a way of changing people's minds. ⁓ facts and figures will never do it. ⁓ if you don't have your facts, you know, somebody can say no, they'll never say yes based on facts. ⁓ you the facts just keep you from saying no. the story that you tell around that that will get people to either believe or not believe in a certain path and journey. that works for both replatforming things like this or just you know, trying to go down and open up a new business line or or anything like that. You can take it small or large. you can tell a good story, you're gonna be able to spread your influence so much more widely than otherwise. And I think people shortchange that a know, I I for a lot of my career I was like, hey, if you just have the right data, you surely people will come along you wonder why they don't. And it's like, well, we're We were like created as a species of sitting around fires, you know, sharing knowledge through stories and that hasn't And that's where people actually decide to do something or to come along for a for a journey. And so if I were to go back to my my early self, you know, I would have said that like learn to tell stories earlier. ⁓ that would have been some of the best career advice I could have given myself.

(36:54) Charles: to to quote Saša Jurić who we just I think we just published his episode this morning, telling a story with your commits and your pull request, all the way through to talking with ⁓ product owners and and managers and decision makers.

(37:01) Mike Ratliff: Very nice. Yeah. That's so so true. I love that.

(37:15) Charles: Speaking of past and and stories, ⁓ we didn't really get into this earlier, but what what's your Elixir story? How did you come to the the language and the beam and

(37:23) Mike Ratliff: Mm-hmm. Yeah. That's a ⁓ yeah, so an interesting I I built the first energy company I worked with, ⁓ we built a platform for demand response and ⁓ in Ruby and Rails. And it was awesome from a like a delivery perspective, like we we were able to build something really quickly and whatnot. it was a two way connected system with all these thermostats and things that were sending telemetry back in real time. And ⁓ that was a It started I as we got as we became more and more successful, it was becoming a real problem. and we were at least struggling to get the the scale to work out when you have these persistent connections, all the always open. It wasn't like you know, have a user logging on and logging off. It was just a very different model. And ⁓ one of the engineers on the team, I can't take any credit for this, that time had a ⁓ put an MQTT server in between those that was written in Erlang. And ⁓ I didn't actually at the time appreciate what that actually meant. It took me ⁓ several years to truly kind of unwind what was really going on there. But ⁓ and then I heard about the very like the very kind of beginnings of the Elixir story starting to unfold and it just seemed like a really ⁓ like the solution that I needed to solve these types of scaling and resiliency problems of the types of apps that I that I was building. so even it was like really early on, like pre-1.0 Phoenix days, and we started we started using it at one of the companies to build a a ⁓ a virtual power plant application. And it was still rough back then. It was probably a you know with the benefit of hindsight it was probably a little early to jump on Elixir bandwagon to do it, but actually ended up, you know, being relatively successful. That company sold to a to another to an acquirer and and whatnot. And I even see the acquirer actually was a sponsor at one of the ElixirConfs for a few years. yeah, that that was kind of the the path and you know, I've just kind of ridden it since then and watched it mature and now it it's almost become a default decision.

(39:44) Charles: What is the so bring it back to the future now, ⁓ but kind of staying with software a little bit, what the future of software really kind of look like in the energy sector and ⁓ and maybe as a second question, ⁓ if you could also tell our audience a little bit about what a virtual power plant is. I think that might not be familiar to everyone here.

(40:06) Mike Ratliff: Yeah, yeah, sorry to use language that ⁓ industry specific. And so you talk about a virtual power plant, ⁓ you know, people different people have different definitions, but I think at its core you're trying to aggregate devices together and to look at how the sum of those devices can be treated as one unified unit. whether they're in one location or even distributed across many locations. kind of casting a an envelope over those and saying these things are going to operate in an orchestration and give you a unified result. And it could be load and generation and storage, know, all mixed together. It could be what we call non-wires alternatives where it's like the behavior of something changes. ⁓ And so all of these things ⁓ get stitched together and you've essentially created a facade around that that you can s like go to a utility or somebody and say, hey You know, we've got, you know, ⁓ a gigawatt and a half of dynamic capacity in your territory. Would you like to purchase that? Or would you like to use that for some event or something like that? And so that's kind of the the the deal behind a a virtual power plant. other question was what's software gonna look like in the utility space? Is that was that what it was? Yeah. Yeah, yeah.

(41:22) Charles: Yeah y or or energy sector, which I suppose might be a little more broad than just utilities.

(41:28) Mike Ratliff: I'm a big believer that our world I think somebody else said this. This wasn't like my observation, so I'm gonna steal it. ⁓ and I can't remember who to give credit for. I think the The world of software, like solutions, is gonna s is gonna saturate pretty quickly. Like the ability to solve software only problems is gonna get easier and easier and faster and faster. it's not like those jobs are gonna go away or anything like that. I'm not a proponent of that level of just think it's gonna be very easy to build when you combine hardware and software or or or software with the real world and like going out and physically touching or changing physical behavior, think that's another frontier out there. And I think the the really impactful software projects are gonna have a lot of those are gonna have a component of that of like how do they reach out and and touch the world. I I think that's that's kind of exciting. ⁓ most of the software engineers that I've worked with that have gotten a taste of controlling something in the real world have absolutely loved I it's frustrating to no end when nothing when something doesn't work and whatnot, but it is also very, very rewarding when it actually plays out. so I I think you're gonna see more and more melding of that and hopefully those lines between those two disciplines will continue to start to blur a little bit more. I would love to to do more and more of those projects.

(42:51) Emma Whamond: We're starting to see a lot of that blurring even now. You know, we've become increasingly dependent on electricity over the past few decades, AI, data centers, electric vehicles, remote work, commerce, moving online. Are we entering a fundamentally different era of energy demands? ⁓ and how is AI playing a role in that? ⁓ what are you seeing on your side?

(43:15) Mike Ratliff: Well, I mean, from a demand perspective, the backlogs of what of how much p power people would like to have access is almost dwarfing how much power we have access to. that's like to put the order of magnitude on on the table there. You you cannot buy ⁓ gas turbines. They're like what, five years out. cannot buy, you know, like You've got people that are that are building bespoke transmission lines outside of the utilities foray and like doing it on a on all like people realize how this is not like a step change, it's like a logarithmic scale change in how much power our world is hungry to consume right now. And so in all this, it clearly was driven, you know, most of it by these these AI data centers, and that's not gonna stop. even with the efficiencies and inference that we see, you know, coming, gonna get dwarfed by the usage of of what's going on. And so we I don't think that most that most people have an idea of the orders of magnitude of power that really need to be acquired. so it's not any one here. It is Yes and to all of the forms of of power acquisition and transmission. And that's one of the biggest problems I think that we have here, especially in the States, is transmission. and why I say that is that ⁓ a lot of the renewable sources are not where are not easily located where we need power and we do not have enough transmission in place to move those ⁓ those renewable sources around our country. if we hope to get to a place where we are more and more reliant on renewables, we're gonna have to figure out a way to build things again and not get bogged down in years and years of permitting red tape that has cost us a lot, I think.

(45:16) Emma Whamond: I completely agree. It's a conversation I've had often here in my my household. boyfriend works in nuclear is a scientist in nuclear. And this has been a huge issue. He just he was working in Europe for a while. and yeah, the energy consumption in general ⁓ just it's it's larger than people realize.

(45:37) Mike Ratliff: Yeah, well. Yeah, I hope we can get back to realizing that there are safe ways to use nuclear because of the new designs of those plants are super promising and ⁓ that is a fantastic way to get relatively cheap and clean and safe power if it's done right.

(45:59) Emma Whamond: what have you been seeing at GridVar about the feasibility of introducing these data centers into the grid? Just give quick rundown. Sorry if it's a big question. ⁓

(46:11) Mike Ratliff: No, it is, but it's a it's a good one. I I love it. I mean it's like that's what we're trying to to figure out every day. know, so obviously like I said before, we're working on the power quality side of this problem. There's a whole real power acquisition side. It's really not our foray is a huge problem. ⁓ we think our our thesis of why we should exist as a company is that there is this power quality issue and the and it's getting only getting worse. it's getting worse because of the level of dynastim of these loads. And people just don't do there's just aren't good models to understand how this is gonna affect the the wider grid. you know, this may not have been a big news story here in the States, but it wasn't too long ago where there was a huge massive blackout in Spain. and when you start to dig into that, is the dynamics of of these ⁓ of things that are happening much more rapidly than they ever did, that that is becoming a more and more regular occurrence. ⁓ whether it's data centers as the as the pure source or other types of things as the source of that, it is just dynastism in general that is that is causing these new problems. so, you know, I'm hopeful that, you know, w these things can be designed in ways that have that take I things are going like this and turn into things that go like this, know, stable instead of chaotic. and that inner if you treat every interconnection point as like it has to have a rigor around its on its operating envelope, I think we stand a chance of evolving our grid. if we let things just behave the way they want to behave, ⁓ and impact outside of their point of interconnection, I'm frightened and I'm glad I have my own power generation source here in my home in Maine. because it it'll get weird if we don't figure that out.

(48:08) Charles: For engineers early in their careers, what lessons have kind of remained true throughout your thirty plus years in in software?

(48:20) Mike Ratliff: I think if you want to be in this industry, you've gotta love being able to learn every single day. And like that has to be just know, that that level of curiosity and it it's not necessarily learning technology. that's the important piece. But you know, pick up those other things, you know, go study some other area but then learn a domain. I think if that was if there's one thing that I can like Tell people it's that no longer living in a world, I I think, where you can effectively be a just a software engineer, be like, hey, tell me what I want to what I should build. days are over. if I'm going to be hiring software engineers, I want them to understand intimately what this domain is all about and be able to make real-time decisions. I I think What's been fascinating for me with like listening to all this rhetoric around on one side of the AI argument that say, God, it's a hundred X improvement in speed, and people on the other side that say, I see no improv I see no speed improvement at all. like I try to like reconcile that because there I know both I know people on both sides of this. I know they're smart. I don't I believe they're experience. what the only thing that I can come down to is that it's this time to pause issue. what I mean by that is if you're getting an AI to write a bunch of code really quickly, but then you come to a moment where a decision has to be made, and you're still working under the assumption where you've got to go talk to somebody, like another group, or you've got to figure something out, it doesn't matter that you got to write the code ten times faster. It's all erased in the no in the process of going back and forth to try to figure out how to move forward. And if you're somebody that really understands a domain, can make that instant decision in that moment and keep moving. And that's what I've seen over the last nine months as I've been building GridVar is like it's not a five or a 10X improvement. It's like a hundred X improvement in throughput. But that's only because there's never a pause. Like when something goes wrong or like there's a question, I can answer it and I can get it back online. that's the that's where all the speed is coming from. Not the fact that the models can actually build code fast. That's nice, but that's a rounding error in terms of all of the air that you take out of the system being able to make these judgment calls. so figure out a way as a if you're a junior engineer, figure out a way to become that person can make those judgment calls real time and get them right of the time. And you're going to be hugely valuable.

(51:02) Emma Whamond: I love that. It's it's so relatable to every aspect of someone's life out outside of just a career device. ⁓ as we get to time and start wrapping up,

(51:07) Mike Ratliff: Yeah.

(51:13) Emma Whamond: where can listeners learn more about your GridVar, or how to get involved in the energy space?

(51:20) Mike Ratliff: Yeah, I mean I think I mean, you know, you can you can check out our website at GridVar dot com. You can reach out to me ⁓ at Mike at GridVar dot com and I'm happy to ⁓ to respond there. and if people want to get involved, ⁓ you know, we're definitely gonna be, you know, looking for Elixir engineers that have a passion and a knowledge around the energy space. And so to talk to folks that are ⁓ that are excited about that.

(51:48) Emma Whamond: Thank you so much, Mike. Yeah. This is a really interesting conversation and ⁓ I've really enjoyed it.

(51:50) Mike Ratliff: thank you. That's been awesome. I appreciate it. Yeah, thanks for all the great questions. I really appreciate it.

(52:01) Charles: Yeah, thank you. I've also ⁓ enjoyed this conversation and ⁓ I look look forward to hearing more about how this space evolves and for our listeners who are not as familiar with a lot of the energy space stuff we've been talking about, while we're an Elixir focus podcast and and that's largely where we're gonna stay, there are a lot of other opportunities out there to learn more about this stuff. The Volts podcast and some others exist ⁓ that are worth listening to as well. ⁓ Our guest has been Mike Ratliff from GridVar, and it's been ⁓ a pleasure to speak with you.