[intro music plays]
Ryan Donovan: Hello everyone and welcome to the Stack Overflow Podcast, a place to talk all things software and technology. I am your host, Ryan Donovan, and I'm joined today by my co-host Eira May. How are you doing today?
Eira May: I'm doing pretty well, Ryan. I am surviving [laughter].
RD: This is the most exciting time of year for us at Stack Overflow. We have the developer survey results out in the wild as of last Tuesday, and we're joined today by Erin Yepis, our senior analyst, who has put together the data, found some interesting insights and is happy to share. So welcome with the show Erin.
Erin Yepis: Thank you for having me.
RD: Of course. So I think, you know, when we get the results, everyone's like, what's new? What's surprising, what's changed? What for you is the most surprising?
EY: I would like to say we try to change as much as possible about the survey this year. So hopefully alot of our regular survey contributors noticed that. So some of the bigger things that we changed were asking questions about AI. AI is changing of course, so we're gonna ask different questions. And we had some different questions, different options, different looks at technology too. So I think the really big one here is that the response we got for trust with AT, the trust scores have been decreasing, which is not something we would have expected. We've only been asking about AI in the developer survey since 2023. We don't have like a long timeline, a longitudinal study going on here, but we have consistently asked the same question about ‘do you trust the accuracy of the AI tools that you use?’ so that has been a consistent question, consistently asked, and the answer this year was a decrease compared to a very steady percentage in the past two years. So it was hovering around like 40 ish percent for trust, not trust. There is like some in the middle, neutral. And this year it's just gone way down. There's way more people that do not trust AI this year compared to last.
RD: How big is that drop in in trust?
EM: Yeah. What kind of gap are we looking at here?
EY: So this year for those that responded that they highly trust or somewhat trust, that came out to 32%. Whereas we were in the 43-45% range for the previous two years.
RD: It's interesting. It almost seems like the more people use AI, the less they trust it. They understand its limitations.
EY: Yeah. I think that's a good hypothesis because we've also been asking in the dev survey if the community is using AI tools in their development. And that usage number is going up. It's higher than it has been since we started asking the question.
RD: I did see a heartening statistic there in the cross tabs. It seems like the people who are most distrustful of AI are the ones who are learning to code.
EM: Interesting, because I feel like that's one of the benefits that's been sort of touted from the beginning is that it, you know, makes the learning curve more manageable for junior developers and people who are learning to code. So if those are the people who are saying like, ‘Ooh, I'm not sure that I have faith in these results’, that is a pretty significant change from what we were seeing the last two years worth of surveys where we asked about AI. Right, Aaron?
EY: Yeah. Again, the hypothesis being that these tools, like a lot of them are in the early stages of being released, and they're learning whoever's putting out those tools, whether they're open source or proprietary, they're learning as they go as well, as well as the people that are using the tools. So I think if you're starting from a baseline that's much lower than someone who has more experience coding, you have more disenchantment with like, I thought this was gonna be magical and save the day, and I'm having a harder time keeping up with the output that's not working.
RD: Yeah, because there have been a ton of articles about how new developers aren't gonna learn how to code because you just have AI or junior developers are gonna fall off. But I think seeing this and seeing how mistrustful they are, where they this doesn't quite work. This doesn't quite get it. I still have to go to Stack Overflow and ask my questions, warms my little heart.
EM: It seems like there was a little bit of a drop in favorability viewing AI tools between 2023 and 2024, but then this year it was just like plummeted. So I guess that's part and parcel of what we're talking about with just our overall sentiment around AI. Maybe not exclusively our feelings of trust, but just do we think it's doing good things for the industry? Do we think it's making developers lives better and easier?
EY: Yeah, and there's definitely lots to dig into there. There's definitely a correlation between trust scores and sentiment scores, so those that typically, most of the respondents that had a lower trust score, also scored the sentiment low as well. But we did notice that there are some bright spots. So regionally there's a couple of countries that had higher sentiment scores for AI tools, namely India and Italy and the Ukraine. Also, I noticed that if we looked at it by age, there is an age group that if you rated AI tools as highly favorable, you also rated it as highly trustworthy. And that was the 25 to 45, those two age groups there. And I think this is something that you called out, Ryan, that sort of self-fulfilling prophecy if you're using these things all the time. It's kind of like, oh yeah, this is great. I love it.
RD: If you already have a high favorability score, like you're gonna use it more. We ask a lot of questions about digging into the reasons behind that sentiment. Do we have any findings on what those frustrations were, what the humans place could be in future AI?
EY: Yeah. We actually specifically asked about frustrations with AI. So the top one that came out was that you're still going to have to ask for help if the AI can't help you with your request, going to human. So I think that gives us another hypothesis that, maybe from the learning perspective as well, spend whatever amount of time using the AI tool, don't get what you expected, don't understand why you didn't get what you expected. You have to double your time by the human interaction part. So that's a hypothesis. Or I mean, I'm sure there are some more like in-depth interviews with actual users. However, we had like a couple of spots in the survey where we allowed users to give us more feedback about AI, which I'm digging into now. But early on in the survey, from day one, we were getting feedback about how I don't wanna read about AI. AI is not as important to me as you might think it is. I think a lot of our hypotheses about what the sentiment and the reasons behind that are. I think there's just a bunch of reasons out there. However, if you're using AI tools for work, we know historically from research on other tools that it's all about saving time. That's why people kind of Stack Overflow like, I'm in a hurry, I need an answer quick, this came up my Google search. So AI is not helping you save time. It's just kind of like rubbing salt in the wound. What am I doing this for?
EM: Right? It's just like an extra step in the process that doesn't give you any new information or insight. It's just, okay, well I got a wrong answer now I still have to go talk to a human.
RD: Yeah. It's interesting that there's some response to the, you know, I become less confident in my problem solving. That's something I've heard as a fear about AI, that you don't learn how to struggle, you don't learn how to solve the problems yourself.
EY: I think that's interesting too. However, the way that there's so many tools out there, regardless of you have a problem with one thing, maybe AI is helping you with it, writing your JavaScript or Python code, but at work, you know, there's problems that AI can help you with, like really basic things, but it's not gonna help you get successfully, get a raise from your boss. I mean, it might help you generate an email where you try to set up a meeting to ask them about it. It's still early stages. It's helping with basic things.
RD: How about in terms of AI agents? Did we ask about the AI agents?
EY: We did a new question this year. AI agents, I feel like towards the end of last year, has been coming up a lot in the news, a lot in like product release information and has been touted as something like everyone's going to be using AI agents at work. So we asked in the developer survey, ‘Hey, are you using AI agents?’ and 31% of users said they are currently using them. So not a majority, not even half.
EM: What do you think accounts for that? I mean, it seemed like maybe this is just sort of seems that way in retrospect, but it seems like people were sort of on generative AI technologies. Like the adoption was sort of like high, pretty rapidly. Right. It doesn't feel like we've seen quite the same thing in AI agents, at least in the workspace. Do you think the survey results are kind of reflecting that?
EY: I think that one, there might be AI agents you work with and you're not even aware of it, so they're so new that you don't even know what to call it and you didn't realize, oh, that's an AI agent. Okay. So that could be part of it. But I think the other part of it is related to the trust scores. Why is a company going to invest a bunch of money into software that they're not completely sure that the security, trust, consistency, workflows, like all of that's going to like make sense. And you still need developers or other tech coworkers to maintain those systems for AI agents as well. I don't know if we have the workforce for that yet. You know, another interesting stat from, or new question we asked and statistic from the dev survey is how many of our respondents had learned to code specifically for AI in the past year? And we saw upwards of 60% of respondents, whether it was for work or for something, a personal project. So those skills still need to be set up for a workforce. They're, if you're have a curriculum put together, like that's brand new. I know there's a lot of courses we've been seeing like out there for different use cases. Very specific usually. But the AI agent stuff I think is still like kind of new and setting up whatever you need at your work or at school to get that to work is still in its like experimental phase.
RD: Yeah. I think if you wanna take advantage of AI agents, you sort of have to do it yourself. Were there any interesting AI tools in the survey? You saw any results from that?
EY: Yeah, so I mean, we asked a lot about AI tools specifically. One big difference from last year, last year we asked about AI search and development tools like and asked which ones users had used and wanna use in the next year? This year we split it up, so we asked specifically, which LLMs did you use in the past year? And then which out of the box agents, so that's where your Chat GBTs and Geminis come into play. And then all the other stuff for like the agents, like the observability, the security, those tools. I think it's interesting, as we saw in the past with, when we asked about AI tools last year chat, Chat GBT was top of the list. So for LLMs we saw that OpenAI’s Model o1 was top of the list, GBT model. Followed somewhat closely by some of the Gemini models. The Gemini models have come up, they were in a couple of questions in the dev survey, and at least for one of the new questions we asked about, which relatively new Stack Overflow tag have you worked with in the past year? That top rated tool there was Google Gemini.
RD: I also saw the AI agent orchestration tools. That's a fairly new one. You know, the early days it was like lang chain, but I was surprised to see Alama up there as the top. I haven't heard much about that.
EY: Oh yeah. I just learned about it myself, but I learned about it from one, it came up in the tags on Stack Overflow, like the relatively new tags, so it was on that list, but I heard about it, I feel like they've done a really great job of 1) putting themselves out there in some of this, like these free courses that you can take. So that is where I heard it from the first, but I think I've been seeing advertisements for them too, like they're growing. You know, a lot of this investment is trying to get your name out there first, so you're like, you have the most name recognition for the specific thing that you do. Not to say I even understand completely what a LAMA does. I pulled it up, I saw how easy it was to like get started. They make it seem very easy to get started on creating your own, your local instance. Yeah. Glad to see that I'm not the only one that's getting a LAMA in the sphere here. There's tons of AI information in the survey, but I know it's not people hear enough about it.
RD: This podcast is for the freshest and hottest takes.
EY: Yeah, that's right. Hot takes only.
RD: That's right. Let's move over to the technology part. I know this is something we've done consistently every year. People are very excited about it. Start with the most popular programming languages. This is the big fight, right? Just looking at it, it sort of matches up to what you expect, right? Like all respondents, JavaScript and H-T-M-L-C-S-S are top of the pops, right? Everybody uses JavaScript like front and back end. Alright, now using JavaScript thanks to node js. The interesting one I thought if you go to the learning to code, is that Python is number one. We had highlighted a discussion that happened on Stack Overflow about what's the best language to start using. And a lot of people said Python 'cause it's easier -
EY: It is consistent with those learning to code. They said the same thing last year. Something I pointed out, I did like a specific cut of the data for Google Technologies, like to see what else those developers are using, if they're using Google Cloud. And Python came in really high for them. And Python is the programming language you use for the Google CLI like the terminal code. I think it's becoming more that Python is being embedded in a lot of stuff just like SQL is, or if you are using Microsoft Azure, then C# is Google is pretty ubiquitous, especially for students. They offer a lot of like free tiers so you can like get your Google CoLab set up and some free like storage space and get started learning on BigQuery for database skills. So, it may be a factor of just like they've made it easier for those learning earlier in their career to like get started.
RD: We'll have a couple follow up podcasts with some Python luminaries, but one of the things I kept hearing is that with Python, you come for the language, but you stay for the community. It's very much a collaborative ecosystem.
EY: At Stock Overflow. We love to hear about collaborative ecosystems.
RD: Indeed. Skipping ahead a little bit, the Rust popularity one, it's interesting to see Rust placement in the popularity contest too. So where did Rust fall this year? Rust is about 14%, which is up from 12% last year in popularity.
EY: And they have that very high admire score too. So those are all of the people that want to work with it, but have not worked with it already. I'm excited because I know the dev survey results get shared out a lot with those Rust subreddits and other sites. So the more that they remain popular in our results, the more that it's going to get shared out and distributed amongst their users. They love seeing that.
RD: So let's talk about databases now too. I think it's interesting, you know, Postgres has had a commanding lead for a while now and it keeps growing. I wonder if you have a take on that.
EY: You know, I don't, because this isn't a company per se, it's like something anyone can spin up. It's open source, I believe, to an extent. But in my working career of working with databases, I've never just worked with a one-off Postgres SQL database personally, my insight there. But you know, I would say otherwise the developer community from other poll surveys that we've run at least recently we ran a poll survey this year for open source AI, and we asked just about sentiment overall, like on open source technology, and so there's high sentiment there. A lot of the more popular tags on On Stack Overflow, including Python, are based with these communities, open source communities or whatever way that they collaborate with each other. They have maintainers. I know Postgres has been part of the survey for a long time. I would have to double check. I don't wanna speak outta turn, but like since the beginning of the survey. So it might just be also that we have a generally older, more mature population of developers that respond to this survey. They have a lot of experience using tools, and Postgres is something that, hey, if it's not broken, don't fix it. It's comfortable maybe, and it's not something you have to pay, like an arm and a leg to maintain with a company for Cloud space and stuff. Well, maybe you do actually, I don't know. Like I said, I haven't specifically worked with it, but it's probably just more attainable for smaller shops.
RD: Well, I've had a couple podcasts about Postgres, and one person said it was the best database for AI. It has a very popular PG vector plugin, and it has a vast plugin ecosystem, so it is very adaptable. Some would say it's the only one you need -
EM: One stop shop -
RD: One stop shop -
EY: Yeah. I wonder how they're going to pivot with all of the MCP server activity going on.
RD: I imagine there'd be some sort of layers above it.
EY: I know Redis is also, they weren't super high on our list for databases, but they were in the top five and I know they have been specifically marketing themselves as we are the AI database. And then the details of which I do not understand completely, but I know that's part of their deal too.
RD: There's a lot of databases marketing themselves as the AI database [laughter], but I know Redis specifically was a caching solution for a long time. I'm not sure people think of it as a database.
EY: Yeah, that's a really good point actually. I think we got a comment when we were soliciting feedback from the meta community about that very same thing about caching. This isn't a database, but -
RD: How do you think they cach things? [Laughter]
EY: Yeah, like I think it's up for debate. I think they are marketing themselves as a database now, so I call it you one, I guess. But the company themselves, I think are fine with that.
RD: So you have this cloud platforms, containerization package matter build to is a grab bag category, right?
EY: Yeah, no, this was a big change this year. In the past we had two different sections where a lot of these tools lived, and I just decided this year, you know, we've seen consistently over the years cloud is how a lot of developers work. It’s kind of getting to be in that ubiquitous stage. So we don't need to separate this out. And especially since I know like you're constantly clicking through the survey, I'm like saving you a couple clicks. I put all of the most popular options and then some into this section and just, you know, this was meant to be like cloud ops technologies, like what are the most popular, you know, the ranking of which makes it a little bit different just because some of the things that used to be on the top of the list for platforms themselves are no longer at the top of the list. You have a long list to look at. A lot of these technologies, there is a long list. We've also gotten feedback from some people that are like, Hey, what if I work with something that isn't the most popular technology for this one area or subject matter? And so, you know, this year we also included like write in your response, like write in what you used, which of course I'm still going through right now. Because one of them, I want people to participate and also I specifically put that at the end of the list so you at least had to scroll through it before you wrote something in. One of my pet peeves with the survey is that people write in responses for things that were on the select list. Like, why are you making me do this extra cleanup? But it's fine. So now that we grouped all of that together, instead of seeing like those top three cloud companies just show up at the top of the list, we have Docker at the top of the list.
RD: Yeah, it's not surprising. I think Docker has in a lot of ways enabled the modern cloud infrastructure.
EY: That's right. And there's some choice that comes with Docker too, right? You don't have to use it, even though they kind of are the biggest name out there and they're easy to use. They're in a lot of the learning materials. If you're setting up something for the first time as well, they make it very easy. Again, no cost, like set up a trial, you can get your own pipeline quick, set it up, connect it to BigQuery. When we had cloud platforms segmented out by themselves before we would see Amazon, Google Cloud, Microsoft Azure show up at the top. When you show up to work, you don't get to choose which one of those you use. Those cost a lot of money. Those are contracts that go into negotiations for months or sometimes over a year. And that's what you, you just have to learn that that is what you get. Whereas I think if you work on a data engineering team, you have a little bit more agency over like, okay, I can set up these pipelines how I want to. Right. And maybe you can choose this over something else.
RD: Yeah. Another thing I thought was interesting is when they combined. Docker is far away the leading one, like in the previous one it was other tools. Docker and NPM were pretty close, but now there's like a 15 percentage point gap between them.
EY: I think that's great. I mean, probably has more to do with the fact that there are, again, like having to work with cloud technology is ubiquitous, so this is something that there are more developers are using.
RD: Another one I wanted to touch on is the web frameworks one. That's one with a really fun long tail, like there's two clear winners there. And then a long tail, it goes all the way down.
EY: There's just a lot of tools out there, isn't there? Yeah. I think that with all of the JavaScript questions we have on Stack Overflow, there's a lot of true web developers that come to the site, part of the community are taking the dev survey. The top developer role that has been recorded is full stack developer. So I think, you know, you're doing a little bit of everything, not just front end work. This is pure like web stuff. So I think it is also like very easy, the nomenclature of things like, oh yeah, I either work with this or I don't. I definitely know what this is. I mean, we've seen over time, it's like a couple years ago, Drupal is really low on this list. But I continue to include Drupal as an option just because 1) we do still have questions about it on Stack Overflow, we have a separate stack exchange site for Drupal, but 2) there are Drupal sites that are important out there. There's Drupal shops out there. It is at the very, very bottom of the list. I wonder, it's been decreasing in usage in the developer survey, at least over the years. I wonder how that's gonna go.
RD: Yeah. One thing I thought was interesting is that jQuery is still number three. That's almost 20 years old. Talking with the guy who created Svel for upcoming podcast and talked about this Cambrian explosion around 10 - 15 years, like all these web frameworks came out. And jQuery is before that, it's still trucking alone.
EY: Lots of Stack overflow questions about jQuery too. That's one of the ones that consistently stays ranked high as well as on the developer survey on Stack Overflow site. Again, I think there's a lot of developers that say ‘hey, why fix what's not broken?’ I know how to use this. It's working great. We need some new ones coming in saying, I wanna try this new guy.
RD: Yeah. Are there other stuff in the technology section that you all wanna chat about?
EY: I made a comment for the IDE section. I spent some time separate from AI, LLMs and AI agents. We have this new sort of incorporation of AI tools into the developer IDE, and it's like, Hey, if you're using this, I don't need to call out one extension of an IDE, like, are you using that? Like, I just wanna know which one you're using. We can get more specific from there. Yeah. VS Code, we've heard anecdotally that VS Code was, is losing their, you know, top market share spot to all these newcomers that are specifically AI first in terms of helping coders to everything you could in VS Code and then some. But those did not rank very high in this list at all. Specifically, we added Cursor, Claude Code, Windsurf, lovable, VS Code and Visual Studio are still top of the list.
RD: It has been number one for a long time. I remember early in my career, people were using VS code. Just got this inertia around it.
EY: I was expecting that the way people are talking about it, that that was gone now, especially considering that Cursor is like a fork of VS Code. So it's using all of the VS Code is, you know, in GitHub you can do stuff with it. No Cursor is like nowhere near used as much as the other IDs we have listed in here.
RD: Cursor is also, you have to pay for it, right?
EY: That is true. That is true. That's a good point.
RD: Yeah.
EM: Yeah. Can't forget that detail.
RD: Can't forget it. [Laughter]
EM: That's definitely gonna have an impact on adoption rates. [Laughter]
RD: I think there's an interesting thing to be said about the admired desire. That's a fairly new part of the survey, and it's interesting that you see a big gap between a lot of the blue numbers and the red numbers, that sort of usage versus appreciation. The spelt creator said sometimes the small languages and frameworks have an advantage because you have to seek it out. So a small niche language has more of a chance to be loved by the people who use it.
EY: Yeah, especially if you get to be the cool person that tells all of your developer friends, like, check out this. Oh, you guys don't know about this cool thing that I know about. And it works really well. Sees felt, got a pretty good admired score too. Nothing to shake a stick at. 62% of developers indicated they want to work with felts, but have not yet.
RD: And you know, that was always the complaint leveled against Rust, which is why it's interesting to see its continued growth.
EM: So just pulling back out a little bit, I'm wondering how do you anticipate that some of these findings in the developer survey will impact the community over the coming year? I mean, how is work and the job market gonna change for developers, do you think based on these findings?
EY: One of the bright spots here is that we saw you can voluntarily tell us something about your salary information and your role. So we saw a lot of roles, their median salaries increased this year compared to last year. So that's gotta be good. You know, if you can get your foot in the door in one of those roles, specifically applied scientists was one, and I think that probably has a lot to do with more of these companies trying to figure out how they're gonna use AI. And again, creating that skillset and that workforce that helps maintain these new tools that are just starting to figure out right now. And then, you know, there's a couple of other roles where we have that salary information in the results. We also have specific comparisons so you can see exactly how much those top roles grew in median salary year over year. So that's good. The job market seems to be pretty stagnant for the technology sector and new grads that don't have any skills or any job experience yet or are trying to get their foot in the door might be a little bit harder, but everyone's learning new skills right now. And one of the big things I think is we asked in the survey, if you have been learning to code for any reason in the past year, that was 67% agreed that they were, regardless of where they were in their career so far. So if everyone's learning, that means there's a reason to learn. If you're going to school or you're just starting out in your career path as a developer, then you're not too far away from the place that more experienced developers are in just because there's still more things to learn. And I think what's gonna help with these AI tools and that sort of dissatisfaction with using the AI tools that are out there now, at least the free ones, is that you have a chance to sort of create that foundational knowledge for yourself. Making it easier so that tool can help you because you know the right questions to ask and you know what right and wrong looks like, and you have just a better sense and intuition about like how to set these things up, what's the right way to set things up, as you gain more experience doing those things hands-on. There's lots of great information in the results that'll give younger people an idea of areas they want to explore. And then just knowing that a lot of people are learning along with them, whether for the first time, like they're still learning how to do stuff for code.
RD: I mean, it seems like there is a lot to learn and every day there's something new with, especially with AI agents and all the MCP stuff. So I know this is a worldwide survey. Do you notice any interesting regional differences? I know you mentioned the greater trust in AI from India and Italy and Ukraine.
EY: One of the top regional differences this year is that Italy made it to the top 10 list. Welcome Italy. So we had to code in a little Italy flag for that graphic, which is very cute. And top 10 meaning like of all the respondents, the most came from these top 10 countries. Regionally, I think we also, we have a breakout for some of the top responding countries on wages as well. So we see some interesting stuff. The US is the top responding country, so there's a lot of responses there, but as we like dig in further to the countries that 1) there's less responses, but also did you provide information for these different questions? I think there's some really good information for the salary profiles and the top roles for Germany and the UK. Which I think, you know, just along with everyone else in last year's survey results, saw that depreciation in their annual salary, take home salary, and now we're seeing that grow there as well, but for different roles, so not just the same ones as in the US. I think again going along with there's applied scientists, like I said, was a big one. That is something that we see specifically in the US. There's a lot of cloud and data engineering going on in those other countries too, and they're seeing the benefits of that increased salary. And yeah, the sentiment for AI, we saw some differences in trust as well, with, other than that, regionally, I haven't noticed anything yet. One thing that is on my list to look at, because I looked at it last year, was job satisfaction by country. Maybe I'll include that in the survey results as I'm working on the code right now. There's a lot of places where I'm like, Ooh, last year we tapped things out so it was a lot of like repetitively looking at professionals versus learners. Like, let's see, more countries, let's see more of the age ranges. Let's just see like a variety of tabbed, cross tabbed results. So job satisfaction. I know like last year, Netherlands came out as being pretty satisfied with their jobs. That's, I think it's gonna be very interesting and probably more of a popular thing to look at for developers and when the survey results are released or as they're released right now. We also asked a new question this year about considering a new role in the past year. And across the board, 45% of the respondents said they were not considering a new role at all. So that probably speaks a little bit. We had overall higher job satisfaction scores and just that 45% number.
RD: Well, it is that time of the show again, I'm gonna shout out somebody who came under Stack Overflow, dropped some knowledge, shared some curiosity, and earned a badge. Today we're shouting out somebody who won a life jacket badge, a junior lifeboat. Somebody came to a question that had a score of negative two or less answer they got a five or more and brought the question up with them. So today we're shouting out Milos for answering a question about one of our high ranked IDEs, sublime Text, three Notepad plus, plus theme. If you're curious about this, find it in the show notes. I am Ryan Donovan. I host the podcast, edit the blog here at Stack Overflow. If you liked what you heard or disliked what you heard, email me at podcast@stackoverflow.com. If you wanna reach out to me directly, you can find me on LinkedIn
EM: and I have been Eira May. I edit the B2B content here at Stack. You can also find me on LinkedIn, or you can reach me just by emailing podcast at Stack Overflow.
EY: And I'm Erin Yepis. I helped put together the developer survey this year and the results site, along with a couple of other treasured coworkers, including Ryan and Eira. And the full results are gonna be at survey.stackoverflow.co/2025
RD: Well, if you're interested, we'll have that in the show notes. Thanks everybody for listening, and we'll talk to you next time.