Tuesday 24 November 2020

Graphistania 2.0 - Episode 11 - The Emil Update

Yey! I got to do it again. For the 4th time in the history of this weird thing called the Graphistania podcast, I have had the change to spend some quality time talking to Emil Eifrem, our fearless leader and CEO of Neo4j. As last time, we actually recorded the video, so you will find the zoom call, and the MP3 version of it, below in the blogpost - along with the habitual transcription.

Hope you will enjoy the chat as much as I did.

Here's the link to the youtube video of the call:

And finally: here's the transcript of our conversation:

RVB:00:00:31.318 hello, everyone. My name is Rik, Rik Van Bruggen from Neo4j. And here we are again, recording Episode 11 already of this year's Graphistania Neo4j podcast. And it's a little bit of a celebration episode because, who do I have on the other side of this Zoom call? Hi, Emil. How are you doing?
EE:00:00:58.913 Hello, Rik Van Bruggen. How are you?
RVB:00:01:01.429 I am excellent. Thank you for joining me. And I have to congratulate you. That's where I'm going to start this, I think. I have to congratulate you because the fourth time on the podcast. You're the only person.
EE:00:01:13.819 Woo hoo! [laughter] That sounds good. I think I am, at most, third most frequent after you and Stefan.
RVB:00:01:22.512 Oh. Stefan would lead, of course, yeah, but he's my co-host these days. Right?
EE:00:01:27.056 Oh. Okay. Sorry. All right. Fine.
RVB:00:01:27.683 Yeah, yeah. So I'm not counting him [inaudible]. No, no. He's part of the furniture. [laughter] It's--
EE:00:01:34.900 Yeah, yeah, exactly. What is he specifically? Is he a table or a chair or?
RVB:00:01:41.541 [laughter] Yeah. Well, I think a couch. [laughter] But--
EE:00:01:46.830 He's very comfy, relaxed. Right?
RVB:00:01:51.170 Very much so. So we have a couple of minutes here to talk, which I know it's been an exciting couple of weeks for you. It's the week after your board meeting. And luckily, there's Thanksgiving around the corner. Right?
EE:00:02:04.500 Yes.
RVB:00:02:04.909 So all that's quite good. But let's start with talking a little bit about this crazy year that we're in.
EE:00:02:14.511 Oh my God.
RVB:00:02:16.164 It's 2020, man. What is [this?]? It's been crazy, huh?
EE:00:02:20.022 It's absolutely crazy. I guess we have six more weeks to go, and who knows what the writers of 2020 have in store for us. But I'm certainly looking forward to turning over into the next chapter.
RVB:00:02:36.712 Yeah. I mean, on so many levels, I think. Right? It's personal level, but also professionally, I mean, all the stuff that's been going on with the world outside. But I think we've doing quite well. I mean, Neo4j, I count ourselves to be the lucky ones. You know what I mean?
EE:00:02:55.305 For sure. Yeah, for sure. Look, I mean, we just talked about that. Right? We can comfortably do our jobs from home, over Zoom, over the keyboard. And there are so many people out there who are in retail or working in healthcare and whatever, truly have to be out there on the front line. So we certainly are the lucky ones. Having said that, I mean, it's still real. The impact is still real, and I'm blown away. By the time we record this, this is literally just a week - I guess it's six days - after we released 4.2, Neo4j 4.2, which was the first 100% COVID release of the database. So this is the first release that was written from couches, from kitchens, from bedrooms, a completely 100% distributed team. And I'm super impressed with how well people have taken to it. In particular, I think the folks who have-- because obviously, we have a wide range of age groups and family situations and whatnot in the company. We're, I think, about 400 people now, right, something like that? But in particular, the people who have kids at home, if you have kids at home and you're homeschooling them and then still be able to actually be a productive member, I think that's super impressive.
RVB:00:04:28.273 Well, we both have kids. Right? And the level of attention they can require and--
EE:00:04:34.904 Yeah. Yeah. I mean, we do, but on my side, Sweden never shut down schools. And so we always had our kids in school and preschool and whatnot. Right? So that's a big difference versus if you have to have them at home, homeschool them, typically on the same network that you have your conference call with a customer or something like that. That's tough. I don't understand how people have been able to do it.
RVB:00:05:02.939 I have a Quality of Service setting on my router. Daddy always wins. [laughter]
EE:00:05:08.442 Daddy always wins. [laughter] [inaudible].
RVB:00:05:13.261 Yes, exactly. No. But yeah, I'm hoping you guys are keeping well and everyone else is. And obviously, although people that have a more difficult situation that we do, that they are keeping well as well. You know what? The reason I invited you back on this podcast is a little bit more specific. Right? I don't know. I think we chatted about it. But it was because of your talk at the NODES conference a couple of weeks ago. First of all, the NODES conference, how good was that?
EE:00:05:43.646 How amazing was that? Yeah. Over 10,000, significantly over 10,000 people registered, thousands of people concurrently watching the sessions. And just the lineup was just amazing. Actually, I think you and Jim talked about it a few episodes ago, maybe it was two episodes ago or something like that, just ahead of the conference. And it was either you or Jim who just looked at the speaker lineup, and it was just like, "Man, this is some really, really, really good content."
RVB:00:06:15.242 Six days of content. [laughter] It was just unbelievable, really. I was very impressed. But I was impressed by your talk as well. I thought it was-- and I'm not sucking up too much here.
EE:00:06:26.966 You're just saying that because, ultimately, I pay the bills.
RVB:00:06:30.851 And I have to be the polite radio show host here. Right?
EE:00:06:33.559 Exactly, yeah. [laughter]


RVB:00:06:35.287 But your talk was interesting because you talked a little bit about the future and how you thought about the future and the next decade in graphs. I'll put the link to the keynote on the transcription of the podcast as well. But I really thought you hit some interesting points there. So maybe we can talk about that a little bit and take our listeners through that. The first thing you talked about was the shift to the cloud, kind of an open door - isn't it? - but an important one.


EE:00:07:08.255 Yes. That totally is an open-- and maybe let me take a step back. So basically, kind of the theme of the keynote was a little bit, hey, it's 2020. Yeah, it's a crap year from any perspective. It also is the last year of a decade or the first year of a new decade, depending on how you want to look at it. Right? And on some level, the previous decade is when the graph category was formed. Right? We walked into that decade with NoSQL having just formed, NoSQL, like in our [back?]. Right? So NoSQL, as a term, was coined in summer of 2009. Right? And so it was just six months before we walked into that decade. And for us, specifically here at Neo4j, our entire focus in the early days of the last decade was make sure that graph databases, as a category, is seen as being part of the broader NoSQL phenomenon. At that time, there was this conception that we disagreed with, but this conception that NoSQL was a product category. Right? It was either SQL or NoSQL. You can hear it in the name. Right? And we always felt that that was wrong. And actually, it was more of a movement or an observation of a shift in how people think about storing data and that several product categories would emerge out of this movement. And one of them, of course, we felt was graph databases.
EE:00:08:35.743 So a huge part of our focus was just making sure that when people talked about alternative databases in general, under the moniker frequently of NoSQL, that they would also consider and appreciate graph databases. And so it's really from zero to one. Right? And that's what broadly happened then in the previous decade. And on some levels, if you really zoom out and remove a lot of the details-- and probably too much of the details. Right? But then my very, very, very 10,000-foot perspective on that decade, what we achieved collectively - we at Neo4j but also the other graph database vendors - is that we made the category happen. Right? It's undeniable now that graph database is a-- it's a formidable force in the world of data. And we have some extraordinary quotes from folks like--
RVB:00:09:34.341 Forrester.
EE:00:09:34.693 --Gartner saying that the future of data and analytics is analysing relationships and data. Right? The future of data and the future of analytics, that's a pretty stark comment. Right? So that's kind of the big one. Then if I look at a little bit more the anatomy, like how did it happen, I think if you simplify it a little bit that the category-- if you look at really the depth of adoption for graph databases in the previous decade, it's some version of it's used for transactional applications. And for that I mean developers building applications, so the direct user of the database is an application that is then used by people, deployed on-prem. Right? That's a little bit-- and it's specifically around certain use cases, things like fraud detection, real-time recommendations, data lineage. Things like that were really the core USP, the unique selling point, or the unique thing about it is we had a significant amount of read queries with multiple hops. Right?
EE:00:10:49.403 And so if you look at kind of-- if you imagine a Venn diagram, it's kind of-- all right, it's transactional applications. It's on-prem. It's deep hop read queries. That's kind of the-- not the only, but that's the bullseye in terms of the real depth of adoption for graph database in the previous decade was that. So that's kind of the starting point for my keynote. And then I said, "All right, so that's how far we've come and how we've come there, if you will. Right? Now let's look forward 10 years. What do I believe from my vantage point, having the privilege seeing most of the adoption of graph databases out there just by Neo4j being a very popular graph database? Where is it going to unfold over the next 10 years?" So that's kind of the broader framing. And then I said, "I believe there are four themes, four big trends that will shape this decade." And the first one, as you alluded to, was the shift to the--
RVB:00:11:53.774 Cloud.
EE:00:11:53.975 --cloud. And we can spend a little bit more time on that, if you want to. And then the second one is the rise of the developer or putting the developer at the front and center for how things unfold in the category. The third one is the rise of data science and, specifically, graph data science. And then the fourth one is my personal favourite, although this is like choose your [laughter]-- like this is the favourite of your children or something like that. Right? And so I love them all equally, but the fourth one is-- I love the fourth one very much, which is, specifically, I think we're going to see the property graph model live up to its full potential. And so those were kind of the four trends that I talked about in the keynote.
RVB:00:12:41.361 Yep. Yep. And the first one being the cloud, and we've already kind of started that journey a couple of years ago. And last year, a big milestone with the release of Aura, which is now expanding, more customers, more services, and all that type of stuff. And that's really the start of our journey on this. Right? It's the start of the decade.
EE:00:13:06.739 That's exactly right. And obviously, we're a part here of a much broader secular shift. It started out 10-plus years ago, and everyone knows that there's this shift to the cloud. What's a little bit more subtle is that the pace at which it's happening is different per layer in the stack. Right? And so it happened faster at applications. Right? It's been probably, I don't know, 20 years since most people started using web-based email applications. Prior to that, it was Outlook, and I used to use Eudora. Do you remember Eudora? [laughter] Right? From [crosstalk].
RVB:00:13:46.867 Showing your age. [laughter]
EE:00:13:48.141 Yeah. And I used Pine and all these kind of things. Right? And then everyone started Gmail and Hotmail and whatnot. Right? And that was probably 20 years or something like that. Right? So that happened very early in the application layer of the stack or on the business side, like CRM with Salesforce and so on and so forth. Right? And then many, many, many layers below in the stack, VM, virtual machines. Right? EC2 was the first cloud computing service from Amazon. Right? And so it happened kind of counter-intuitively at the very top of the stack and almost at the bottom of the stack. Right? And then different layers from the stack had moved at different pace. And it took data, frankly, longer than I thought. You've been around for a long time, owing to your extraordinarily old age. [laughter] You're welcome. But also in the company, so you probably remember the really early-- like in the early days of Heroku. I want to say 2012, 2013 is when we wrote our first cloud service, right, Neo4j as a service. And then we've kind of been dabbling and haven't really been able to properly focus on it. And then in '17 and '18, it hit an elbow in the curve when, all of a sudden, the data layer of stack really-- things started shifting. And it was held back by two things primarily, data gravity and compliance and regulatory reasons. But both of them were kind of overcome, and we hit this tipping point a couple of years ago. And that's when we also said, "All right. Now it's time to go all in on this." And so we released Aura, which is this cornerstone for us when it comes to the cloud in 2020.
RVB:00:15:27.254 I was really amazed by the speed that it evolved. I mean, three years ago, I remember writing emails to you and Lars saying, "This is not happening in my customers." [laughter] And it went way faster than I thought it would. But what's the second one? The second topic was all around developers.
EE:00:15:48.020 Around developers.


RVB:00:15:48.561 This is also a topic dear to your heart.
EE:00:15:52.729 Very near and dear to my heart and very much where we grew up. And I guess I'll take-- the very zoom-out point of view on this one is that I think that the internet has changed consumer behaviour in general. Right? And it comes back to this big business model innovation that happened a gazillion bazillion years ago, actually, in grocery shopping. Right? And this is in the '30s and the '40s or something like that, 1930s and '40s, a long time ago. Right? When it used to be that you went to the grocery store and you asked the shopkeeper and you said, "Give me four eggs and give me a bottle of milk." And they would look up and--
RVB:00:16:37.929 They would [crosstalk] behind the counter. Yeah, exactly.
EE:00:16:40.030 Exactly. [laughter] Put it in the cart and stuff like that. Right? And then there's this one guy who realised that, "Wait. We can have consumers do this themselves." Self-service has been the term for that business model. Right? They can go into-- and then they go to the cashier, and they check out on the way out. Right? And so they can speak to a store representative whenever they want if they need help, to find something or whatever, but they don't have to. There's a bunch of stuff that they can do where they don't require to wait in line for someone who's employed by the store to help them out. Right? And so that is what was the innovation. That created supermarkets. Right? And so the internet really enabled that in so many ways for consumer behaviour across the board, right, including for things that traditionally have been very enterprise software-centric, things like infrastructure software. Right? And for when it comes to specifically developers-- and I say this with all kinds of love and infatuation because I'm obviously, as you know, a developer myself, or probably by now an ex-developer. I don't know if I can write production-- well, I know that I can't production code anymore. But that's [crosstalk]--
RVB:00:18:05.605 Let's hope. [laughter]
EE:00:18:07.622 Exactly. Well, I certainly don't. For those of you who are waiting to download 4.2, no worries, none of my code is in that release. [laughter] But there's a lot of stuff where a developer-- where they don't want to talk to a human being in order to try things out. Right? And we have this rise of self-serve, API-based-type services, where you can opt in to the full product experience. You can learn about it at your own pace with your own material. You can evaluate it, you can try it out, you can sign up, you can use, you can expand, and you can ultimately cancel if you don't want to keep using the service, all without having to speak to someone. But, very importantly, with an option to do it when you feel like, "No, now I actually need to engage with-- I want to engage. I need coaching [inaudible] a more strategic, broader discussion where I want to engage with another human being." Right? And that is a huge part of what I see as the future for graph databases in this decade, a self-service way of consuming the database.


RVB:00:19:20.993 Interesting. Right? And I think this is something that, like you said, it's been on the move, and it's been evolving that way for quite some time. But it's interesting to see it move into the traditional realm of enterprise software as well. [inaudible] of time, and I know that your airpods are going to run out of steam, [laughter] so we probably need to get going here. The third one that you talked about, Emil, was all about graph data science. You did a couple of-- well, actually, during your keynote, Alicia was doing a little bit of a segue tour of our GDSL, Graph Data Science Library. Right? But to me, this has always been a fantastic topic because Neo4j has always been great at these pattern-matching queries. But what if you don't know the pattern? Right? [laughter] Right? That's kind of the basic view that I've always took at this. How do I find the pattern? Well, now we actually have a pattern-finding engine of some sorts. You know what I mean?
EE:00:20:25.715 Yeah. Yeah. That's exactly right. And I think that maybe the way the perspective-- [I'd lead?] into it. If you heard what I said at the beginning, at the top, just framing the entire thing, about the previous decade, I said the depth of the adoption, not all of it, but the depth was this Venn diagram of on-prem and then applications, so developers writing applications where you have transactional workloads is how you typically talk about it in the database world. Right? And this really is about-- so the first trend we talked about was expanding on the on-prem thing. Right? The second point is about expanding on that second thing around purely transactional applications, where all of a sudden, we're seeing that there's this obviously huge rise, much like the cloud. The fact that there's going to be a rise of data scientists and AI and machine learning in the world should come as a shock to absolutely zero people listening in to this podcast.
EE:00:21:24.215 But it's actually, if you look at how that field has evolved over the past-- basically since it's birth, it's very similar to the world of databases, pre-graph databases, which is-- it's been founded on, "Hey, there's a lot of valuable stuff you can do with tabular view." Right? But you leave a lot of data untouched or unseen. My favourite analogy for that is you can look at the world from a tabular perspective, but it's a little bit like looking at it in black and white. You see a lot of valuable things, but there's an entire dimension that you don't see. Right? And then if you start looking at relationships and data, how things are connected, all of a sudden, you see in colour, or maybe go from 2D to 3D is another analogy. Right? And on some level, in the previous decade, we achieved that for transactional applications. And I think in the 2020s, that is what we're going to do for the world of machine learning and data science and artificial intelligence.
EE:00:22:27.799 And this comes back to the first ever GraphConnect that we ran. And for the listeners, GraphConnect is our-- it used be annual, but then something happened in 2020. But it used to be our annual conference. It will soon again be our annual conference, where we bring everyone together in one location, back when that was safe. Right? And we talked about what's going on in the world of graph databases. And the first one, we had a professor from UCSD, University of California in San Diego. And he spoke about-- he'd written a book called Connected, which, of course, is the type of name that we tend to love here at Neo4j. It was Professor James Fowler, and his observation was that-- he had some really, really stark research results, where he was one of the first people to get access to the Facebook social graph in order to do research on it. And he had uncovered some interesting things. For example, he was able to predict voter turnout with a statistically significant much higher degree of certainty if he knew the graph around the individual, and just two hops out even. He could predict that with a higher degree of certainty than if he knew everything about the individual.
EE:00:23:50.071 So in other words, if he doesn't know anything about Rik Van Bruggen, name, age, nothing like that - he just knows Rik's graph one hop out or even two hops out - he can predict with a high degree of certainty whether Rik will turn out to vote in the election and even things like whether Rik is a smoker - those are the two things that he had researched - than if he knows everything about Rik: medical history, lifestyle, everything about you. Right? And so the fact that those relationships around you are such an important signal in order to do predictive analytics-- which really is what a lot of machine learning and data science comes back to. It's like, "Am I able to predict the future from existing history?" That's a really huge component, and it's completely ignored today. If you don't know how to operate and connect the data, then you can't use that as a signal. And so that we obviously see a huge amount of usage of, we call it, the Graph Data Science Library, GDSL, and Alicia demoed it as part of our keynote. And we see a huge amount of use of that already, and I think that's going to be a big growth driver for the world of graph databases in the 2020s.
RVB:00:25:08.462 Totally. Yeah. There's a lot of stuff in there in the-- I always think about it. If you look at the graph first, it's so difficult to know where you start. Where do you start looking for interesting data? Right? And the GDSL library or some of the ranking algorithms that are in there and stuff like that and clustering algorithms, they give you pointers. They give you, like, "This is where you should probably look a little bit more. I'm not going to tell exactly what it is that is going to pique your interest or that's going to be meaningful to you, but this is probably where you need to look more." And then there's a lot of-- and actually being able to filter like that is super, super valuable to a lot of people. Right? So [crosstalk].
EE:00:25:52.749 It's so valuable. And I think the Google analogy is probably applicable, where it's like, "Hey, look. You type in your search query, and I'll tell you the top 10 search-- I can't tell you for sure that it's going to be in one of them. But out of the entire web, let me pop the bubble to the top the things that you should start investigating more." Right?
RVB:00:26:14.875 Exactly.
EE:00:26:14.966 And of course, as we know, what they use for that is eigenvector centrality or PageRank, which is a graph algorithm.
RVB:00:26:22.825 I vividly remember the days of AltaVista and Lycos and those shitty, shitty search engines--
EE:00:26:30.953 Exactly. [laughter]
RVB:00:26:31.395 --and my uncle telling me, "Rik, you should really try Google." And I was like, "Huh."
EE:00:26:36.803 Really? Wow. Well done, Uncle.
RVB:00:26:38.815 That's how it happened. Yeah. My uncle, he pointed me to it. Yeah. It was a--
EE:00:26:42.574 Either your uncle is an amazing early adopter or you were just a really, really late adopter. [laughter] It could be one of the two.
RVB:00:26:50.429 We're going to go the last theme. [laughter]
EE:00:26:52.836 Let's move off this.


RVB:00:26:54.103 Yes, move off of this topic, dammit. And that's a good finale, I would say, the full potential of the property graph model. And I think I know what you mean with that, but it's always been a little bit weird for me. Why don't you explain it to our listeners, what do you mean by that?
EE:00:27:14.766 Yeah, yeah. So again, I'm going to start-- my jumping off point will be how I framed the previous decade, where it's primarily on-prem, and then applications, transactional workloads applications. And then the third one I said was driven primarily by read performance, specifically around multiple hop queries. Right? And graph databases are magical in that the type of performance that you can get compared to other database models, when you want to jump across connections, when you want to figure out how things are connected, it's just extraordinary. That's when you get the-- it's a thousand times faster. It's a million times faster. It's just these mind-blowing numbers. Right? And so a huge amount of the adoption, in particular, the commercial adoption that we've seen for graph databases, have been for that, where you just can't solve it any other way. Your alternative is either you try to solve it yourself, which is basically I tried to do my own in-memory graph database-- either that or I don't solve the problem. Right? Those are typically the alternatives. You can't go to a relational database or a document database or anything like that. Right? Huge strength of ours. Right?
EE:00:28:31.356 Having said that, I've always believed that that's just one of the three core value propositions of a graph database. And I believe the three core value propositions, if you peel away everything else, is intuitiveness, agility, and speed. Those are the three ones. Right? And this truly just talks about the third one, speed. And what the observation that we've always made internally, but certainly I've always made, is that the graph model gives you a richer vocabulary to express most data. And you can most easily look at it when you look at the document database model, which is, [of course?], a hugely popular database model out there, where you talk about JSON documents and how do you serialise them into a database. Right? And you look at that, and you can capture these JSON key/value pairs. Right?
EE:00:29:29.334 But all of a sudden, if you're going to say that, "Actually, I'm storing people and cars. I'm doing some kind of a vehicle register," or something like that, then you want to be able to store vehicles, and you want to be able to stores cars-- sorry, vehicles, and you want to be able to store people. Right? And that's very easily expressed in a document database. But you then also want to connect them. Right? You want to say that, "This car has an owner called Rik." And all of a sudden, then you have to have some kind of a field in Rik saying, "All the cars that I own, all the vehicles that I own." And you have to have some kind of licence plate, unique ID, or a VIN number, and you have to store that. It's basically a foreign key is how we would think about it in the relational database model. Right? And that you can't really express well in a document database. Right? You can express it, but all of a sudden, when that car gets towed and destroyed, that VIN number disappears. Then all of a sudden, your pointer in that document points to nothing. And then you have to compensate with that in your code. "Hey, if nothing gets returned back, then it's probably been destroyed." Or you have remember when I destroy it, I have to look at all the documents that might have encoded a pointer to this one, which there's no real support for. Right?
EE:00:30:54.090 And you can't say things like, "Hey, when was this owned by me?" Well, you can easily attach information to those relationships because really what you're doing is you're trying to encode relationships in the real world in the form of documents. Right?And this is something that is entirely doable. Right? The document data model is isomorphic with the graph data model, just a fancy way of saying that you can express all data in these various different forms without losing information. So the question isn't, "Can you express it?" The question is, "How easy is it?" And the observation that we made early on, going back 10, 15-plus years when we sat down and we first conceived of the property graph data model, was that actually most domains are connected. And what I used in the keynote, in the notes keynote, is I brought up an incognito - so it's not polluted by my search history, right? - incognito Chrome window, went to Google image search, and I searched for domain model. And every single visual, every single picture on that slide was a deeply connected domain model. I went to Wikipedia, the article for domain model. You bring that up, it's a big graph. It happens to be a healthcare model. It's really, really complex. So you understand that it's the US healthcare system? [laughter] And then you look at that, and you say, like, "Well, if most domain models look like this, why are we choosing to store them in a format that doesn't really appreciate and easily represent connections?" Right?
RVB:00:32:37.495 It's a compromise. Right?
EE:00:32:39.044 Exactly. You have to compromise. And I think the answer for that is that because the relational database exists and it's really mature and it's an amazing piece of technology-- because document databases, many of them are really easy to use. And I think, let's be candid with ourselves, we have a lot to improve there. I think Neo4j is-- it's the most easy-to-use graph database on the planet, but it's not yet as easy to use as it could be. Right? And I think that there's a bunch of applications out there that have a thousand objects or a thousand records or elements, however you want to think about it, where if you have as easy of a surface to the property graph model as you have to the document model or to the relational model, it is a better fit, and it is much more productive to get up and running with it. Right? But we haven't really talked about that as much. In fact, maybe we talked about it 10, 12, 15 years ago, in the really early days. But more recently, like in the past 10 years, we've been more focused on, what are the core use cases where we are--
RVB:00:33:42.147 [Of graph?], yeah.
EE:00:33:42.403 --a thousand times faster, a million times faster? There's no subjectiveness to it. It's just undeniable. We are CPU clockwise just so much faster than everyone else, which has been the low-hanging fruit, so to speak. And I think that's been an appropriate focus of ours. But as we really start growing up as a company and as a category, I believe that there's an opportunity that we have. And it's not guaranteed to pan out, but there's an opportunity. It's a moonshot-type thing, where if we manage to make our product as easy to use as humanly possible-- and I talk about moving it from this usability hierarchy, all the way from impossible to possible to normal and then to magical, what I call magical, where it's just so easy to use that it's a delight. If we do that-- and I believe that since the core data model, the core property graph data model, is the most widely applicable one for most applications, I think we have this opportunity to become the first database where for most new projects, people end up reaching for the graph database, reaching hopefully for us, Neo4j, but more importantly for a graph database as that first database of choice. And that--
RVB:00:35:05.171 And that's--
EE:00:35:05.419 --I think, is an extraordinary opportunity for us in the graph space in this decade.
RVB:00:35:11.316 And that's the full potential, as you [crosstalk].
EE:00:35:13.864 That's the full potential piece.
RVB:00:35:15.683 Yep. Super. Hey, just one more question. We've talked in 2015. We've talked in 2016. We've talked in 2018. And we've talked in 2020. When are we talking next? [laughter]
EE:00:35:28.938 I certainly hope in 2021.
RVB:00:35:32.524 On the podcast, I mean. [laughter]
EE:00:35:36.608 Well--
RVB:00:35:37.275 [Not ever?]?
EE:00:35:37.490 Hopefully, more frequently than that without being recorded.
RVB:00:35:41.524 Exactly.
EE:00:35:42.365 But yeah, maybe let's do an end-of-the-year type of a tradition of it. Right? We should meet-- whenever you start thinking about Santa, you should also think, "Hey, I should talk to my CEO."
RVB:00:35:55.828 [laughter] Yeah, exactly. Sounds like a plan. Emil, thank you so much for taking the time. I know it's a busy time of the year and everything around it. So thank you so much for doing that.
EE:00:36:09.222 Awesome.
RVB:00:36:09.270 I really appreciate it, and I look forward to next year at the latest.
EE:00:36:14.292 Awesome. Thanks, Rik. Thanks, everyone, for listening in.
RVB:00:36:16.121 Thank you. Bye.
EE:00:36:18.354 Bye.
Subscribing to the podcast is easy: just add the rss feed, find the show on Spotify, or add us in iTunes! Hope you'll enjoy it!

All the best


PS how did you like the new tune?

No comments:

Post a Comment