1. Library
  2. Podcasts
  3. To Be Continuous
  4. Ep. #53, A Closer Look at Open Source Today
To Be Continuous
38 MIN

Ep. #53, A Closer Look at Open Source Today

light mode
about the episode

In episode 53 of To Be Continuous, Paul and Edith discuss the evolution of open source over the last 20 years, maintainer sustainability, and the frequency of miscommunications in the world of texts and emojis.


Paul Biggar: Today we're going to talk about open source.

Edith Harbaugh: Yes we are, Paul.

Paul: That was odd.

Edith: It's been a while.

Paul: Yeah, it has. Oh my God. How long since we put out the last one?

Edith: Let's not even talk about that.

Paul: OK. Yeah. We're here to talk about open source.

Edith: I think the Valencia bus line was still running last time we recorded.

Paul: Jesus Christ. The big things that bring open source to mind is, apart from the extended thought pieces on open source that it seems everyone is publishing.

I've even got a draft. There's the Amazon stuff, what did you refer to that as? Strip mining? Was that the word you used?

Edith: It's not my word, it's what other people are calling it.

Basically there's some very popular open source projects and Amazon is coming along with those as a service.

Paul: Right.

Edith: Basically taking all the work and in the minds of some, not contributing back.

Paul: This is Elastic and Kafka and Mongo, and there's one more. Redis.

Edith: Yes.

Paul: There's probably other companies that are going through similar concepts, but these were companies that were largely open core.

Edith: Yes.

Paul: There are also companies that have varying degrees of relationship to the product that they're selling.

The people who own Redis Labs, they aren't the people who made the technology, whereas in the case of some of these like Mongo, the company actually created the technology.

Edith: Yeah.

Paul: There's a certain amount of some of these companies are putting the investment in, the current investments, and some of them also put in the initial investment.

Now the cloud providers are coming along.

Edith: Not all cloud providers, it's almost exclusively Amazon.

Paul: Is it?

Edith: Who else?

Paul: I mean, I guess it kind of doesn't matter, right? Amazon is so big.

Amazon is coming along and Amazon made an actual Redis thing.

Edith: Yeah.

Paul: Then it made a Mongo compatible thing. Then it made an Elastic distribution, which is essentially a fork.

Edith: Yes.

Paul: We don't know what happened with Kafka. Something happened with Kafka.

Edith: It metamorphed.

Paul: No.

Edith: It woke up one morning--

Paul: As a shiny new butterfly.

Edith: As a gigantic bug.

Paul: So, you feel Amazon is-- What?

Edith: I don't know if I feel that, but I know some people's opinion is that there are these open source communities that a lot of work has gone into, and Amazon is coming along and strip mining.

Extracting the value and not contributing back.

Paul: It's interesting. If they were contributing back, is that good enough?

Edith: I don't think it is. I think the root cause-- We talked about this in our earlier episode about two or three years ago with Nadia, open source really hasn't found a business model.

It causes a lot of angst, because usually contributors are like "I'm contributing all this. Why isn't anybody paying me?" On the other hand, their developers are like "It's open source. Why would I pay them?"

Paul: Nadia is coming out with a new report. Remember she had that report two, three years ago.

She's got a new one that's coming out at the end of the summer. I ran into her at the weekend and she was telling me about it.

She had some interesting positions that I don't remember enough to quote.

Edith: This is going to be a great episode.

Paul: "Someone, somewhere had an opinion."

Edith: "This company that we referenced, but we can't remember what they're doing."

Paul: This brought a lot of these issues to my mind.

I've been thinking a lot about open source, and about sustainability of products, and "How do you make your product sustainable in the long term?"

How open source is one way to do this.

Edith: I think open source reaches a breaking point. I think capitalism is the best proven model for making a project sustainable.

I'll state upfront what I mean. The business model where you or a customer have a problem and you pay a company to provide a solution, it provides a really direct relationship--

Paul: We say this obviously as startup founders, as people who are able to realize a lot of the value of their creations that we make.

Edith: I've seen many friends in the dev tool space who've run into trouble because they think that the money the VCs give them is to fund their development without having to then find it for the customer.

Paul: I feel one of my skill sets in life is convincing VCs to pay for my fund compiler projects.

Edith: It's worked out so far.

Paul: So far.

Edith: I don't want to name names but, friends from Heavybit who are like "I have this cool project."

Paul: Right. It gets turned into a company and perhaps a company that struggles with a business model.

Edith: They can get a seed round because they are some hot shot out of Twitter.

Without thinking too hard I could think of at least four companies who got easy $3-6 million dollar seed rounds.

Then it's like "Well, I just thought I got funded."

Paul: What you're talking about is that open source has various business models.

Edith: Does it?

Paul: There is the open core, there's services. People don't like services. People have been gravitating towards the open core.

Edith: Well, services doesn't provide VC-level returns. I started off as a consultant, it's a body game. It's like you pay somebody--

I was a fresh graduate, I got paid the equivalent of-- I got paid $60K a year.

I got billed out at $300 an hour and they made the money on the spread, but there's no multiple there.

Paul: What is the Kubernetes company?

Edith: Heptio?

Paul: Heptio just got acquired for $600 million?

Edith: They didn't publicly release it, but I heard it was in that range.

Paul: They were a company of relatively small single-digit million in income, but what they were they were the hot shit in the hot space.

VMware needed the hot shit in the hot space, so they got venture returns for what I understood to be a services company.

Edith: I think they got the money and Joe Beda, who was a CTO and had started Kubernetes.

There was basically, I'd say, a very expensive acqua-hire. They weren't doing it for the revenue. They were doing it for--

Paul: $600 million, though.

Edith: If you're VMware and you're about to get left behind in the container wars, it's like Facebook buying Instagram.

Paul: Right, or Microsoft buying GitHub.

Edith: Yeah. It's like, "The alternative is to get left behind. This might seem expensive, but nobody wants to be the Sun Microsystems."

Paul: Yeah. The services model, I agree, doesn't have the VC returns.

Edith: I'm not saying that it's bad.

Paul: No, but I agree with you. In fact, that was the point I was getting at.

The model that remains is open core and Amazon has just taken an ice pick to open core.

Edith: Let's go back for a second to something that I get asked a lot, is bootstrapping versus VC money.

Idon't think there's any value judgment that people throw around the word "lifestyle business" pejoratively, but if you're making a nice services business and making 40% returns every year, that's awesome.

Paul: I am extremely pro-whatever business model is your favorite, however, I will add I think it comes in here.

I think that doesn't come up when people talk about lifestyle businesses.

I'm going to talk about bootstrap businesses, which is less pejorative, versus VC-backed businesses.

Certain companies have business models that work for those companies.

You're often looking for this fit between founders and business model, but there is also a fit between product and business model.

Edith: Yes.

Paul: A very obvious thing is Dark could not have been built yet without VC.

We're biting off so much that you can't really do it as a lifestyle business.

Edith: It's funny. So LaunchDarkly, our company, has been around for about 5 years now.

Paul: Do you like my LaunchDarkly T-shirt?

Edith: I'm so happy you wore it again. It's getting a little frayed and maybe I should give you a new one.

Paul: Yeah, I think so.

Edith: One of our first 10 customers, the product manager left and he wants to start an API business, so he wanted advice from me.

He asked about bootstrapping versus taking VC money. For us, it was a year before we had anything sellable.

That was a year after we had an idea, we had a concept. We were talking to people who wanted to use it.

We had people in early beta, but it took us a year. I don't think we could have bootstrapped.

Paul: With open source, a lot of these businesses are bootstrapped on another company's dime by open source.

They are at this point, possibly, where they've got tens of thousands, maybe even millions of installs of some kind before the company is founded.

Edith: I think people sometimes confuse installs with customers.

Paul: Yes, I think so.

Edith: We talked about this when Armin was on with Glenn Solomon. People think "Open source, you just go immediately viral."

I mean, Armin was talking about how they got 100 downloads in their first year. Like, a year.

Paul: But again, that was something where Mitchell was this hot shot because he had made Vagrant.

Then I think he spent quite some time building up a brand. I don't think it was HashiCorp's brand, but his own brand before HashiCorp.

Even then, they still had to spend time and building the usage up.

Edith: Yeah. It wasn't an instant hit.

Paul: Right. MongoDB went through a similar thing. It really worked at this thing and made, for all of its faults, which are many and I'm still bitter--

But it brought a lot of innovation to the database game, most specifically that usability matters.

Edith: I think that's the most underrated thing in open source.

Paul: Mongo did all this work in bringing databases, and there is a need for companies that innovate or for people who innovate to see returns on that innovation.

Whether it's paying the VCs back or, in open source, a lot of the innovation pays off in some way. Perhaps you're employed to work on it.

Perhaps you are the high status, perhaps companies want early experts in this particular technology that's going to be hot and you got into it early so now you get all the cool consulting jobs.

There's payoffs of some kind, and the payoff that infrastructure and database companies have tended to see is the payoff from essentially reselling compute, and that's going away.

Edith: The original genesis of a lot of open source was a revulsion against, at the time, the monopoly practices of Microsoft.

"We have 3,000 engineers and you just have to pay us the license fee."

A lot of open source was this revulsion of, "We will assemble 3,000 engineers from all your competitors and build something else."

Now, like you said, that succeeded to a certain point and now people are taking those assets and using them for business purposes.

Paul: Right. There's been a co-option of the freewheeling idealist open source/free software foundations of the what, the 80s and 90s?

Edith: Yeah. It was very much against Microsoft, and now Microsoft itself has done a 180 and is a huge fan of open source.

Paul: Right. Yeah. I remember in the early 2000s was my first experience with open source stuff, and Microsoft was the enemy.

Edith: Yeah. I mean, you were at Mozilla.

Paul: Exactly.

Edith: You were, "IE must die."

Paul: Yeah. I was also "Windows must die." I'm a Debian user. In 2002, I guess, was my first Debian install. The first issue in Ubuntu--

Edith: You were using Debian back then?

Paul: "Using" is a strong word. I did most of my development in college on Sun Machine, SSHing via Putty on the Windows Lab machines.

Which I think is what we all did at the time, basically. A couple of years later, everyone was using Linux machines.

We didn't have home internet, so my lovely Linux machine wasn't really that usable with its not-internet.

Edith: You didn't have home internet?

Paul: I went to college in '99 to 2003. There wasn't really-- I don't know if there was even a phone line, what would it--

Edith: Stifling so many--

Paul: I know. I worked in tech support for an internet service provider and I didn't have home internet.

I lived next to the college, so I spent all my time in college. What would I do with home internet?

Edith: I remember I would go to the computer lab because I did have a computer in my dorm room, but it just wasn't as nice as the ones in the lab.

Paul: We literally spent all of college hanging around the lab.

Edith: The kids don't do that anymore.

Paul: Don't they? I mean, that makes sense.

Edith: They're all on their, I don't know, snap goggles doing VR programming?

Paul: Oh my god, we're so old.

Edith: The whole idea of staying up all night in a lab. I think people don't do that anymore.

Paul: I know. I've slept in the lab multiple times.

Edith: Yeah, I remember. I would stay up all night and then take a cat nap between 6:00 to 8:00, then go to class.

Paul: That makes sense. Back then, open source was really about freedom.

People had various opinions and fought all the fucking time on Slashdot about which version of freedom was the correct version of freedom.

I feel that in open source today, there's almost no discussion of that. The Free Software Foundation version of freedom is gone, essentially.

In retrospect, they were probably more correct than anyone. Exactly what they predicted would happen, is what happened. But no one cares.

Because what people really wanted-- There was a blog post today that Steve Knabnik, the RUST dev evangelist, formerly, talks about "What is open source?"

What a lot of people want from open source is the ability to use it and the ability to produce things.

Where we ended up with open source is it is a community more than anything, in my opinion. It's a way of getting yourself status in the community. People think of status as maybe a negative thing, but I mean it more in the fundamental human condition sense.

Edith: I'd say also in the mid-2000s, I was at Vignette, which was a content management platform.

At the time, we had a whole staff of solution engineers just to do POCs.

You would do a three month, six month POC, which is basically to install the software and show the customer that it even installed.

This is in the hell of supported platform matrix, like CC and what version of Oracle, and even what version of your apps.

Paul: Right. Oh, my God. Bad times.

Edith: Yeah. I managed a supportive platform matrix.

Paul: People keep asking "Why does Darker matter? " You need to remind them of 2006.

Edith: Yeah. It's like, "OK, what flavor of Sybase that this one huge client uses works with which version of Linux?"

Open source started taking away a lot of our business. It's just that people could install it on their own. They didn't have to get solution engineers on site, they could just get open source content manager and just start using it.

It wasn't that it was necessarily better, it's just like they were like "I don't want to have to work through these 6 month hell cycles."

Paul: Here's the thing, that today there are things that are easier than that.

Edith: I'd say that's why SaaS came along, because open source is great. It's like the developer is like "No longer I am beholden to these jerk SEs. I could just get it myself."

Paul: I think this is a fundamental thing about usability, the installability is the thing that matters.

Open source, to so many people is just a distribution channel. It's a way of getting your software into the hands of people.

Then there's multiple different ways of approaching it. If you're Amazon, you already have that.

You already have this giant thing and people can use it. Amazon products aren't exactly the most usable, but they're certainly accessible.

Edith: I'd say the APIs have taken and I'd say improved on what people wanted from open source.

A lot of why people liked open source as a consumer was A) that they weren't beholden to a tech giant and B) that they could get it.

Paul: The beholden matters, I think.

Edith: Well, so C) I'd say there's a whole API economy now of Twilio, and Stripe where it's like "OK, I don't have to install anything. I just get easy access in a second."

Paul: I definitely have opinions here.

Edith: Yeah. I would say that's the same root itch that open source was originally.

It was "I don't have to go through some huge IT cycle. I can just start using it."

I'd say that even looking around the Heavybit clubhouse , "Read me," "Instant APIs."

Paul: I think that may be the downside of the API-ification and how it relates to open source, is that people feel very strongly that they want to be able to take what it is they are doing with the person they're paying and do it somewhere else if something goes wrong.

If the person pisses them off, vendor lock-in is exactly what I'm talking about. APIs don't provide this.

When you're looking at the AWS products, it goes both directions. People are locked into Elastic search.

Amazon, therefore, needs to provide Elastic search, because otherwise people can't move to the cloud and hit Jeff Bezos' dream of being our king, finally.

Edith: I thought he already was.

Paul: Mostly.

Edith: Yeah, it's his world. We're just shopping in it.

Paul: APIs make it easy, but Amazon also makes it easy, or cloud providers, or SaaS providers also make it very easy to do these things that have been hard to do before.

Edith: I think the reason why some people liked open source was this community-driven thing.

The reason why other people liked it was this perception that it was free and easy to use.

Now Amazon has basically picked up the easy to use. I think, at the end of day, a lot of people do not want to install and run basic functionality.

We've talked about this before, I have this theory that value moves further and further up the stack.

People in fact do not want to really run a lot of core services.

Paul: Right. This is interesting. When we look at what open source is, we've established it's not the same as it was in our youths.

Edith: I am still in my youth.

Paul: Community is a large part of what it means to be open source.

Lack of vendor lock-in is another one. The distribution channel is another reason.

The things that we haven't really talked about that are related to all this is the downsides from the whole thing, the maintainer sustainability--

Edith: Something you didn't explicitly mention is the idea of being able to tinker.

Paul: OK, right.

Edith: Which I think actually happens less and less now.

Paul: Yes, that's very interesting.

Edith: One of the original ideas of open source was that you as a contributor could, the old saying, "Show up with code."

Paul: Yeah. What is it called? The nuclear option. The fourth right, or whatever it is, is being able to fork. I think that is actually gone.

You pretty much can't fork things for societal reasons more than anything else, because in order to fork--

Let's say someone suggested recently that we should fork Elm because Elm made the decision moving from 18 to 19 that that didn't work for us, but we were never going to fork Elm, That would be ludicrous.

Because then we would have to build this other community.

We'd be doing it without the thing that makes Elm "Elm." Instead people are given the choice.

You have the choice of being on and off the train, you can contribute and you can maybe affect the direction.

When you look at why large companies are paying so much money for people to work in open source, it's so that they can control the direction of the train. The forking is gone.

You look at GitHub and you see mountains of issues where people are complaining or making issues, but aren't fixing.

I'm not actually sure how much tinkering is going on.

Edith: I'd say that was the original dream.

Paul: A million eyeballs makes bugs shallow?

Edith: Yeah, I think there's fundamental issues about how open source is structured that makes it hard for people to tinker.

Paul: Do you mean that community dynamic, or something else?

Edith: I think this is a fork that we should abandon.

Paul: No, I think this is actually good because I think it relates to the fundamental changing of open source.

Because back in the day, the 90s, all that it took to make an open source project that people wanted to use was that you put it on Freshmeat.

You distributed the tar ball, and then people would show up and use it or not use it.

What's different today is that the ecosystem has other needs.

We're drowning in software that solves the problem that we're looking for.

The things that we need are documentation, and usability, and things like "How do we actually run this thing? How does it get produced and security updates and all this other stuff?"

There is a need for a lot more leadership and a lot more other intangibles that aren't just code.

Edith: Yeah, my pointy-headed boss viewpoint is that the idea of open source is basically "Screw the rest of the work, all we need is some really bright developers."

Paul: That is a position that I used to hold as a developer.

Edith: You know, the "Bullshit walks, show up with code. We don't need this product manager, UE designer, QA, or dock person."

Paul: Yeah, "We don't need salespeople, we don't need marketing."

Edith: "We sure as heck don't need HR or finance."

Paul: Oh my god. The HR one is interesting, because I think open source needs HR.

I think there's a lot of lessons needed in playing nice with others.

Edith: Yeah. It devolved into "We'll show up with code or not." You have unusable UIs.

You have two people that don't get along and no way to resolve it.

Paul: When we think of open source leaders, we're thinking of Linus, who's an asshole.

Stallman who's creepy, and ESR who's fucking crazy town over there.

We didn't have good role models for this community that we built for ourselves, is what I'm saying.

As a result, perhaps many other contributing factors, we have this place that when you perhaps look at the values of the community, you're like "It's not the most enticing place."

Edith: Yeah. I mean, there's been the famous wars with Note about whether they even needed a code of conduct.

Paul: We went through some sort of code of conduct thing recently where there was, I don't want to name the project, but a community was forming around a couple of projects and someone made the inevitable "Let's have a code of conduct." and someone responded by like--

Edith: "We don't need no stinking--"

Paul: No, writing their own code of conduct on the fly to avoid the one that everyone else is established on.

Edith: It's funny because I have two conflicting worldviews. One is my high school had this saying, "Verbum Sap Sat."

Paul: What does that mean?

Edith: A word to the wise is sufficient.

Paul: Yeah, that's not how humans work. Like an honor code bullshit.

Edith: My secondary school was from 12 year olds up and you could go off campus. They didn't take attendance.

You could call your teachers by their first name. You picked your own classes.

Paul: OK. Yeah, that sounds pretty good.

Edith: We had about a 30% complete failure rate of people who just could not handle that at all.

Paul: Yeah. That's real.

Edith: Because if you can go off campus all the time, some people are like "I'm going to get some slushies from 7-11."

Paul: Slushies. That's adorable. I was assuming that they would go off and do drugs, which is what happened in my high school.

Edith: I was not cool enough to be in that crowd. You had kids like me who were like "OK, I can take a ton of classes and graduate when I am 16."

Paul: You finished high school when you were 16?

Edith: Yeah.

Paul: Holy shit, Edith.

Edith: Yeah. I was valedictorian.

Paul: Very impressive.

Edith: Then we had people who were just into slushies.

Paul: Why did you point at me when you said that?

Edith: Well, because you said it was adorable.

Paul: OK.

Edith: I used to think people didn't need a code of conduct. The older I am, I'm like, "Yes, you do."

Paul: Yeah. Just from running companies, you start to see that it takes a lot of extremely explicit work to have a good culture and to have a good vibe and to just--

For everyone who are in this room to work well together.

Edith: Just stuff you assume people know. Like when we got our first interns in, one of them would make a huge mess in the kitchen, then just leave it.

It turned out that he had been at high school or college and when you made a mess it's like, "The janitor will come clean it up."

We're like "No. We have a cleaning person come to do overall deep clean, but if you spill some coffee, you need to grab some paper towels and clean up."

Paul: There's a metaphor there for open source.

Edith: Well, just some basic stuff like "OK, this is on us to train these people."

Paul: Right. Training college grads to be humans. It's what our companies do.

Edith: Just some basic stuff you just assume people know and they don't.

Paul: I find that the biggest one in open source is that people don't know how to communicate to other people, especially over the medium of text.

Which is the unfortunate medium that is chosen by all of them.

Edith: Tell me more.

Paul: You send someone an email, they send you an email, you send it back. Suddenly you--

Edith: Are in a flame war.

Paul: Yeah. You hate each other, you don't even know this person and you read their words with a different way than they intended their words to be read, because text is fundamentally loss-y and you don't know very much about these humans.

You don't know how to read the things that they say in the way that they intended, so you read it in whatever you assume based on your own life experiences.

Then you have a dysfunctional community where people are being assholes to each other.

All of them think it's everyone else's fault and it is everyone else's fault, but it's also their fault because just humans are hard.

Edith: My rule, if there's two exchanges which seem hostile, I stop emailing and I'm like "We need to get on the phone."

Paul: Oh yeah, that's good.

Edith: Even better is in person, but it's just like "Something is getting dropped here."

Paul: I have started to assume that everything is a misunderstanding.

Edith: "Somebody is being misunderstood, let's get on the phone and sort this out."

Paul: Yeah, I think I've done the same thing. So many times I just write "I think we must be misunderstanding each other.

Because the thing that you say is batshit crazy. Therefore, there is probably a set of worldviews that we are approaching this problem from in this conversation from that leads us to have different interpretations."

Edith: There could be so many things going on. It could be that they just found out that their dog is sick, or their mom is having surgery, or they didn't sleep much and they typed it in a hurry.

Paul: Right. They type out a single word and it came up as much more aggressive than they intended.

Edith: They try to use an emoticon and it went completely flat.

Paul: Right. They used the emoji from the Apple character set, but it displayed to their messengee on the Microsoft character set and the smiley face with a slight grimace had a much more aggressive tone than they had intended it to have.

Edith: Or they thought they were being sarcastic and cheeky and it just went completely wrong.

I think open source is still looking for product market fit.

Paul: It's had a hard pivot from the Free Software Foundation version of it to the OSI version of us. I think has undergone a slow pivot to more of a commons.

I think what we're seeing is the open core method of getting value out of open source contributions is going away.

Edith: Even beyond that, we touched very briefly on it, but the burn out by the popular single person or two person projects.

Where they're like "I'm doing this because I love software. Why aren't I getting paid?"

Paul: There's a couple of companies that are dealing with this, or that are trying to solve this problem.

There's Tidelift and there's DevFlight. Have you heard of either of them?

Edith: Yeah.

Paul: There appears to be some sort of backlash against Tidelift that I didn't follow.

Edith: There is an extremely thriving business in cataloging open source projects that a company uses.

Paul: Yes. One would assume that this could then be used to funnel money to them in some method.

Edith: But no, they're just like "We're making money. Why give it back?"

Paul: Yes. There's this problem of sustainability and the maintainers want to be paid.

There's almost a unionizing of open source going on.

I think fundamentally the thing that we have been calling open source in the form that it's been in for, I'm going to say the last three years or so, I think everyone's just coming to the realization that it's broken.

Edith: Yeah, it's the old penny gap problem. Have you heard of this?

Paul: No.

Edith: If people are used to something being free, it is extremely hard to get them to pay something.

Paul: I think that's only one of the problems.

Edith: Also, the most famous thing is Amazon, they sold so much more when they did free shipping.

In France, for legislative reasons they had to charge something, so they charged a penny.

They had this really interesting AB test where they could literally see what their sales have been before and after.

It was a huge drop when they started having to charge a penny, even though it was just a penny. It's just perceptions of their value changed.

Paul: On the relationship to open source?

Edith: People are used to paying free.

If you're used to something being free, it's very hard to get people to start paying for it.

That's what parking is struggling with.

Paul: Right. OK, we recognize that that open source is broken.

One of the ways that people are trying to make people pay for it is saying that maintainers deserve to be paid. You're saying yes, but companies don't have to.

Edith: Yeah.

Paul: It's just like there's no one--

Edith: I'm not saying that they shouldn't. I'm not making a moral judgment. I'm just saying that--

Paul: I wrote a blog post about this that went absolutely nowhere, no one read it. The point of it was that open source needs salespeople.

Edith: Well, and they also need procurement and sales.

Paul: It needs a thing where a human goes to a company and says "You are using this software, and this software needs money, and you have some existential risk, and you need to send money its way or you are going to have some sort of existential risk."

Edith: Then you get into the non-technical side of the business which says "OK, can you fill out this security questionnaire? Are you GDPR compliant?"

Paul: Right. Then you're like "Well, there's all these concerns that I have, so I may as well run it on Amazon.

I'm not going to trust this elastic.co. Who's heard of this company? I want the one that Amazon runs."

Edith: This gets back into product market, which I think there's a fundamental mismatch. I'm not saying which is right or wrong between how--

I'll pick a fictitious-- German Railway wants to buy software, versus how people want to sell it.

Paul: Right. Because people want to write code.

Edith: Yeah. And then German Railway says "Are you compliant with this regulation? Have you passed this sock thing?"

Paul: Are you saying that there is a need for companies to exist in order to solve these problems?

Edith: I'm not sure. I'm just saying the more I sell software, the more I realize that there's a lot more to successful software selling than just coding.

Paul: I mean with CircleCI, we saw there was a ton of open source competitors.

The value that we provided to companies partially was that they didn't have to use them.

Even when we had competitors who were open source, in a lot of cases people were not factoring that in because they wanted the thing that solved their problem the best, not necessarily the thing that had the highest minded ideals.

Edith: So they want to know that you'll be in business?

Paul: Right. Yeah. There's nothing like giving someone money to help them stay in business.


It's funny because the original idea of open source was based around that concern that the company would go out of business and you wanted to have the source code.

This was the heyday of you used to have to escrow code.

Because there was the idea that "If for whatever reason you go out of business, we want to be able to take the code back."

Paul: I think that was true in a couple of cases.

I think what has changed a little bit today is this recognition that if the company goes out of business, you are not going to be able to run the software that they made.

Edith: No, if AWS went out of business--

Paul: Right. Yeah. AWS could open source everything down to their microcontrollers and there isn't the ability of someone to take that and run it.

Even LaunchDarkly, which is how many people? 80, 100?

Edith: 100-ish.

Paul: 100-ish. All right, so let's say 50% of that aren't in direct software roles or development roles. Presumably, it is hard to run the thing that you make.

Edith: Yeah. We have knockoffs all the time now. It's funny because they just pop up and they're like "We're cheap and easy." It's like "Well--"

Paul: It's hard to make this software. It's hard to scale it, it's hard to keep it operational.

Edith: Yeah. We do 200 billion features a day.

Paul: Even if you're a team which is only using a handful of the subset of features that you made, if you open sourced your product, there would be someone trying to run it and not really having any ability to keep that software going.

Edith: It is hard. What we really are is a real time database, which is massively distributed.

Paul: We have the same problem with Dark. The biggest value of Dark is that we run the infrastructure.

Running infrastructure is hard, but is also much easier than creating software for other people to run on their infrastructure reliably.

Which is extremely-- Ten times as hard, I'm going to say.

I think our main conclusion from this whole thing, we probably need to finish this up. Open source is broken.

Edith: Open source is still looking for product market fit.

Paul: I like my formulation better.

Edith: I think there is something there.

I think open source is successful, but it's still looking for a successful way to monetize itself so that maintainers don't get burnt out and companies are not doing abuse of the commons.