Same. I've very wary of investing or getting involved with any company that has an open source strategy. I worked for one (Sendmail) and it did not go well. We talked about how it's hard to compete when your number one competitor is your own software for $0. If you look at successful public companies with open source, they are all fairly small successes compared to their public closed source peers.
When I worked at reddit, we open sourced our code. It didn't gain us much. We got a few contributions, and we also got a few competitors. The biggest benefit was being able to show people the code when they suspected shenanigans.
At Netflix our ethos was "if it's involved in infrastructure, open source it, if it's related to movies, keep it closed." That actually worked pretty well. But the open source code was difficult to use unless you invested in the entire ecosystem. The main benefit was recruiting.
For my own company, I try to support open source by donating to the people who make it.
Same here. I run a small game company in San Francisco, and we constantly are donating to help fund projects like Blender and Godot because they're amazing projects that deserve funding to keep them afloat.
But would I invest in an open source company? Eh, that's a different story. I love them for what they are, it's amazing they exist, but I don't see them as an investment that pays you back dividends. Rather, I see them as a useful tool you can make money with and is worthy of your donations.
> "if it's involved in infrastructure, open source it, if it's related to movies, keep it closed."
In the first wave(s) of open source, I think most contributors were working on company time, but the open source was not the primary product of the company. It was something useful to the business of the company, without being the business of the company.
So what you describe seems to fit into that model.
That model seems to be less and less of open source contribution though. (I am not certain why). And I think that has a lot to do with current apparent crises in open source sustainability -- it was the model that worked, for a while, to produce and maintain lots of open source. I am not sure "open source companies" ever really have.
I don't think that model has gone away. That's largely the basis for how all those large companies (FAANG's included) contribute to open source, and some smaller companies as well. The model is IMHO fairly well explained by "sharing the maintenance burden of commodified infrastructure" in some cases, and in other cases by "commodify your complement".
Open sourcing the core value proposition of your company? Yeah, that's where all those OSS startups are struggling, and then get a lot of flak for shifting to proprietary licensing (e.g. Mongo, Elastic).
Ah, so in general useful for helping customer service. I could see that being helpful for many developer focused services, but definitely for the reddit community I bet it assuaged some concerns.
The author makes a good case against open source businesses, but he indulges himself in a false dichotomy between a traditional proprietary model and open source.
Increasingly often, there is no market for your proprietary software. It is then not a choice between open source and proprietary models, but rather between doing open source or doing nothing.
A lot of markets are just gone, and they're not coming back. Odds are, you're never going to sell another proprietary UNIX clone, or a proprietary database, or a proprietary application server. There are no punters. Those niches have been extinguished.
The market for new proprietary databases is both large and lucrative -- it is printing money right now. What has changed is the structure of the market and the product economics. The classic marketing-driven retail database product model, the kind that puts billboards on the 101, stopped working as a financial model a decade ago -- development costs are too high for the marginal return. That has been replaced by models that produce much higher margins with much lower development cost. Unlike database companies of yore such as Oracle, they use little overt marketing because they don't sell through the public. It is a completely different value proposition for the customer too. And this market is growing.
Proprietary databases have found their niche with qualitatively higher performance and throughput -- often 10-100x -- than the best open source has to offer on the same hardware footprint. As data continues to grow exponentially and cloud costs increase, operational efficiency has a much bigger impact on the bottom line than licensing cost. This gap has only been increasing over time. For most open source databases this is a fundamental architectural issue, and you can't fix architecture, so they are defenseless against highly optimized competitors.
I was on a project a year or two ago where a particular ElasticSearch workload was replaced by a proprietary solution that demonstrably offered 100x higher throughput. It was apparently worth tens of millions of dollars to the customer for that improvement.
The reason big retail databases went away is that 90% of the development cost is in the "big retail" part and not the "database" part. At the same time, development costs for a state-of-the-art database engine have become extremely high due to wage costs (worse than FAANG), so the industry needed to find a business model that would allow those costs to be contained. It is still around but much more invisible. It would be nearly impossible to properly finance a database company as a startup, so a different model is used.
> Proprietary databases have found their niche with qualitatively higher performance and throughput -- often 10-100x -- than the best open source has to offer on the same hardware footprint.
This is not true in general. It depends on the workload. Here's an example: MySQL 8.0 is benchmarked at over 1M QPS on I/O bound reads using Optane SSDs. [1] You would need to show that proprietary software hits 10-100M QPS for this workload and hardware configuration. Oracle and SQL Server are often faster because they have better query planning but it's not orders of magnitude.
This really isn't comparing against Oracle or SQL Server, which are both legacy architectures. And you don't need Optane SSDs; people do better than that with cheap NVMe JBODs like you get on EC2. Also, something is only "I/O bound" if it is literally running at the theoretical limits of the storage hardware. And even that doesn't imply another database couldn't drive much more workload with the same theoretical bandwidth; many popular databases employ architectures that unnecessarily burn storage bandwidth.
It is pretty rare for these days for a modern database engine to be I/O bound in a storage sense, even with cheap solid-state storage. Certainly not for fast-twitch query workloads. For most data models and workloads, the bottleneck is either memory or network bandwidth if you are doing everything else right.
MySQL isn't typically held up as a paragon of query performance. It is solidly middle of the road. It is part of a generation of architectures that can't get anywhere close to current high-performance architectures. Every database architecture suffers this fate as they get older.
There is a Stonebraker quote somewhere (can't find now) that basically says 'give me the workload and I will build you an optimised database'. The point with building a database enterprise is to build something that is sufficiently general purpose - not just optimised for one workload even if that workload is valuable for one company. In that world, there is only open source. Even snowflake, which is cited as a good example of a new proprietary, is built on FoundationDB and is like a hosted version with lots of bells and whistles. We are building a DB (TerminusDB) and started as prop and we couldn’t see a way to make it work without going open source.
This is essentially correct. Trying to be all things to all people massively increased development costs for diminishing returns, and not infrequently forces significant architectural tradeoffs too.
The new databases tend to be narrower, hyper-optimized data model and workload specialists. It requires a small fraction of the code and you know exactly who your target market is. I much prefer this model from an engineering perspective.
It is also what drives the interest in database meta-programming scaffolds that can codegen narrowly optimized database engines from something that approximates a specification. It greatly reduces cost/time to market in this new environment.
I'm not sure how you would open source a codegen implementation. I don't use any open source inside high-performance implementations myself for good reason, though I occasionally do PostgreSQL mods for companies for fun. Customers care less about open source on the very high-end because vanishingly few people could credibly maintain it in any case.
Snowflake is based on FoundationDB which was proprietary for years but is now open sourced, developed by Apple. So it's maybe more like open core than totally proprietary.
This is a really good point. However, there are still successes that I'd never have expected. Jetbrains is a good example. Who would have thought that an IDE could be sold when there are adequate ones for free?
There are also many many smaller niches where software can be sold because there isn't adequate developer interest to create OSS. One example is this: https://querix.com/products/lycia (not a customer, just ran across their product).
I personally have been being using Intellij for everything for the last five years.
I tried to migrate to VSCode for golang.
I mean, don't get me wrong, the situation is not the worst possible, but the only reason you would want to use that is because you never tried refactoring with Goland.
Are you kidding me? I had to go look for what is the current needed extensions for go that are still maintained, configure them etc...vs just download and use Goland.
If you can't or don't want to spend money OK, but otherwise is a non sense to use anything else.
I want the computer to do as much work as it possibly can. My brain only runs at 0.0000001 GHz because it’s made of meat.
But I want to use a concise, powerful language that requires more work from the compiler, rather than an impoverished language that assumes an IDE will generate a lot of crud that isn’t worth reviewing.
I did, there's never been really good community driven open source IDE's. For every community driven IDE I can name you an enterprise driven IDE that's popular and arguably better.
Code editors are a product where polish has high value, and users are willing to pay for that polish. I don't particularly care if my editor is open source as long as it's got strong community and great polish.
IntelliJ is open source. It was once proprietary but these days is "open core".
The Ultimate edition is the open source version of IJ with more plugins. However the rebranded non-Java versions like Goland aren't open source, unless you consider the IJ base. They are the core language plugin(s) + integration and rebranding code.
From what little I've tried it, the open-source Object Pascal IDE Lazarus doesn't really feel all that much worse than Bo– ah, In– eh, Co– uh, Id– erm, Embarcadero Delphi.
Yep IntelliJ is great. I remember back in the days, changing from Microsoft visual studio with resharper to eclipse. Hurrrr what a pain. Then you'd try something else, have to reconfigure everything, never happy.
Then I discovered the makers of resharper made an IDE which supports tons of languages and never looked back.
As long as they don't fuck it up like Adobe fucked photoshop (and every product they own), they'll have my money.
Exactly, and open source companies make great money too. ElasticSearch being a prime example, yes AWS eats a large share of their pie, but who cares, the pie is absolutely huge.
Another example is GitHub and Gitlab. Both got big by leveraging the open source community for marketing. Imagine all the resources GitHub spends to enable open source, that's a cost center that will never go away. If they hadn't done it though, they simply would never have become big.
And then Gitlab one ups them, sacrificing an even larger share of the pie, allowing free private repos and free on site deployments. And still they're carving up a big chunk of pie for themselves.
In my opinion it's the only way the future is going to make any sense. It's not a race to the bottom, it's a race to more balance between the rewards of beating the network effect.
If you want to build a new company in a space that another company is already offering significant value, then you are going to have to offer even more value. And if there's no more value go squeeze out of your functionality, and you're already not charging your users, that means you're going to have to give part of your company to the users. It's a mathematical inevitability.
You should know, Github offers free private repositories. I believe because Microsoft doesn't need Github revenue, but to use it as a funnel towards Azure.
Is GitLab profitable? They seem to have $120M in 2019 revenue but I'm not sure how big they are.
GitHub got big but then sold to MS. I'm not sure it was ever big in terms of profits. If the argument for investing in open source companies is that they can get a big exit by selling to a proprietary software company then I'm not sure the argument is strong.
> The author makes a good case against open source businesses, but he indulges himself in a false dichotomy between a traditional proprietary model and open source.
More generally people seem to get tunnel vision about open source models. The big distinction in business software is not open vs. proprietary but SaaS vs. non-SaaS, with SaaS increasingly dominant. OSS companies therefore need to have a SaaS strategy. You build your own on community OSS projects, supply other SaaS businesses, or use OSS as an on-ramp to your SaaS product.
Well you could start with a SaaS-first business model. People are obsessing about Amazon, but the fact is if you publish open source software and don't build an outstanding SaaS for it, somebody else will do it for you. It can't be an afterthought.
Second, add IP on top that makes the service easy to develop applications for, secure, easy to integrate with other SaaS services, etc. Amazon gives you a service but for the most part they don't build ecosystems.
Third, make it work on multiple platforms, not just AWS.
Fourth, integrate with AWS Marketplace and get their reps working for you.
for tooling or infrastructure or languages this seems true, but most plain SAAS is proprietary, almost all end-user applications are and so on. I'm not entirely sure what percentage of the market that is but it's probably the largest one?
> A lot of markets are just gone, and they're not coming back. Odds are, you're never going to sell another proprietary UNIX clone, or a proprietary database, or a proprietary application server.
I would argue this encourages innovation in freedom-respecting/open-source software business models.
Oracle, Microsoft and SAP all do very well out of their proprietary databases for OLTP. In analytics, Snowflake was the largest tech IPO ever last year. For whatever reason, the database market seems to be entirely ok with proprietary software.
Well, how many companies are there that are selling some sort of proprietary software and still making money? I'd wager there are hundreds of thousands of them around the world. Also, software can now be sold as a service, so the market is just transforming, no?
If engineers are the most important resource, I think being a company known for open source might act as an inbound sales for attracting top talent, too (see: Google, RedHat, Mozilla).
For engineers, it makes sense to work for companies that do have FOSS projects because they can always point to their corpus of work when switching between jobs as proof of their worth. Most engineers who realise this don't ever want to go back to working full time on closed source projects.
That said, folks making arguments against open source companies conveniently ignore the examples of Cloudera, Linaro, Hortonworks. They also tend to forget how much central open source is in the current climate because someone or the other will attempt to commoditize your core (Firebase vs Supabase, GitHub vs GitLab, Oracle vs Postgres, Docker vs Mesos, Windows vs Linux, Intel/AMD vs ARM, S3 vs MinIO, Symbian/iOS vs Android, Plaid vs Moov.io) and because of the Interwebs, these FOSS solutions will be found and a community will develop around it. It is almost inevitable.
I'm not an investor, but one should definitely invest in FOSS companies with proven tech because they attract the most precious commodity of all, developer mindshare. I'm sure FOSS companies will discover newer ways to make their businesses work (like GitLab and HashiCorp with "buyer-based open core"), and blueprints from the past are already available (especially, in terms of what not to do; looking at you, Docker). Despite what they may say about BigTech stealing their thunder away... I think it is mostly down to FOSS companies digging themselves into a hole.
Source-available licensing is a compelling alternative, but I hope the community rejects it, because such licenses don't seek to truly benefit the overall software ecosystem rather only merely appear to do so, and that's why I hate those.
There's possibly "companies with FOSS products" that arne't "FOSS companies".
For instance, I know people that work for heroku that work on open source (and are primary maintainers in some case) on company time -- but heroku is obviously not a FOSS company.
Or, in a comment below jedberg says 'At Netflix our ethos was "if it's involved in infrastructure, open source it, if it's related to movies, keep it closed.' -- nobody would call netflix a "FOSS company".
In the original wave of open source, this was the position of most contributors to open source I think. Work on a thing on company time that helps the company achieve it's goals/business, but is not the main focus of the company.
This seems to be less and less the case. Because companies are more ruthlessly focused on the bottom line, and don't think they can afford to have anyone working on something they don't own or that might help a competitor? I dunno.
But this change has a lot to do with the current apparent sustainability problems in open source.
> This seems to be less and less the case. Because companies are more ruthlessly focused on the bottom line, and don't think they can afford to have anyone working on something they don't own or that might help a competitor? I dunno.
For one, if one's open source product is so good that BigTech wraps its software around it, then that's a cause for validation and celebration. For example, it is pointless for the Rust community to bemoan Google's use of Rust to improve ChromeOS' security; or for Dr. D. Richard Hipp to bemoan pretty much every company for building money making software around SQLite.
Besides, like Bryan Cantrill points out in the blog post I linked to from above, it is often trivial for BigTech to expend effort to re-implement the API surface (MongoDB -> DocumentDB) or even the rewrite whole software (MariaDB -> Aurora) if that's what they need to do to compete with anyone's FOSS / source-available offering.
> There's possibly "companies with FOSS products" that arne't "FOSS companies".
Of course, I am not saying FOSS is the one true way when just recently tech companies like Zoom and Snowflake have demonstrated the potential of proprietary tech; but just that FOSS will commodotize your core, may be half-heartedly, or even half as effectively, but it will happen and there are business opportunities to be built around it. In fact, FOSS is one of the last few avenues left to take on BigTech, as far as technology is concerned.
When I say FOSS, I don't mean the kind of throw-all-commits-over-the-wall, superficial-community-engagement FOSS Google does with Android and Chrome. Google can do so because they have got the money to keep funding "an open source loss leader". For any other upstart it'd be suicidal to keep such an iron-grip on their FOSSd codebase (like Elastic will find out in the coming years). The right approach for such upstarts is to create a Foundation around the tech (k8s) or donate to one (Cloudera / Hortonworks), and generally try to really rake in the benefits that being open-source brings.
If we are cherrypicking examples, here's one from the time when Cisco reluctantly opensourced a router's firmware because GPL, "Linksys continues to earn millions of dollars per year selling an 11-year-old product without ever changing its specs or design... That product really is what made the company." https://thenewstack.io/the-open-source-lesson-of-the-linksys...
Yes, open source gives you developer mindshare, and cheaper hiring, but both of those mean squat if you can't monetize well (and most open source companies I've seen don't).
Let's face it, a tech company's most important resource is one of two things:
* its IP
* the data it's collected on its users
The first precludes open source within the company's core competencies. The second makes the company a place where talented, morally decent engineers wouldn't want to work.
$6B, It’s not so far behind[1] and that before exiting, and I don’t know how much founders retain vs GitHub founders.
But what if would be less, not everything has to be about market valuations.
This is unusually insightful for a LinkedIn post. One question, though: towards the end, it suggests that GPL/AGPL products, which are more defensible against commercial competitors who simply resell the open source project, are "legally murky" and might be rejected by customer legal teams. But does that matter? Can't you just do what Sleepycat did, and offer commercial customers a clean commercial license?
> One question, though: towards the end, it suggests that GPL/AGPL products, which are more defensible against commercial competitors who simply resell the open source project, are "legally murky" and might be rejected by customer legal teams. But does that matter? Can't you just do what Sleepycat did, and offer commercial customers a clean commercial license?
It isn't an issue of "clean"; GPL/AGPL aren't "legally murky". (Though if a customer is willing to pay you because they think otherwise, by all means.) But if they're "rejected by customer legal teams", that's an opportunity to sell an alternative license or exception. I've collaborated with the legal teams at various companies that review both inbound and outbound FOSS. Inbound, there's no open legal question with GPL-family licenses; they're only "rejected" in cases where they're functioning as intended and that isn't desired. (That includes companies that systematically reject AGPL; they're concerned about the license working as intended.) Outbound, the pressure to not use GPL-family licenses when a choice is available doesn't come from the legal department; it comes from product teams who don't seriously consider copyleft and what benefit they might gain from using it, because all the other teams they see are using Apache or MIT or BSD. And then those same teams become shocked when their own code is used to compete with them.
There's a strong tendency towards "do what other people are doing", and there are a lot of companies releasing permissively licensed code. This seems self-defeating.
> It isn't an issue of "clean"; GPL/AGPL aren't "legally murky".
Yes and no. I'd agree that, in general, they're legally safe and clear, but when you're talking to the legal department of a company, it is not about whether they're clearly defined. The actual worry is whether a patent troll or a more litigious competitor can sue the s out of your company and extract millions upon millions of either legal costs or settlements. Yes, you _might_ win, but it will still cost you and courts are known to not always be deterministic in their decisions. So once your out of the "thrown out without a second thought"-cleanliness, corporate legal will not be happy.
It really depends on what you're selling. All of the Linux core tools and the kernel are released under GPL and obviously companies have no issues using them. For libraries that are statically compiled or used as part of your own software it's a different story though.
> For libraries that are statically compiled or used as part of your own software it's a different story though.
That's still not "legally murky"; that's the GPL working as designed. If the component has enough value, that's an opportunity to sell an alternate license.
Of course, if your primary value proposition is something other than the library (such as a paid service that the library helps programs integrate with), by all means use a permissive license.
Microsoft spent a lot of money and energy in the late 90s and 00s to convince the world - especially business and legal people - that the GPL was a "cancer"
The GPL is one of the cleanest licenses out there. Use the code in your project, give the source of your project to anyone you give the binaries to, job done.
A lot of people wrongly assume that if you license the product under GNU GPL, you must publish source code on the internet. Many active proponents of free (as in freedom) software had very tense quarrells with me when I explained them that it's not exactly mandatory.
This is a small point to argue. Even if you choose to distribute your source on-demand by CD in an envelope with postage stamps on it, the next logical step is for a customer to just put that code on github. So, what then is the difference?
Putting it on github (or a zip file on your website) is less of a hassle.
If the customer paid a really high price for software and source code (like a few million dollars), he might rightfully decide that publishing it and giving it to everyone for free is not in their best interests.
That used to be the business model of Adacore and the GNAT compiler.
Their compiler was technically GPL, but they sold it for thousands of dollars to businesses, with the implied threat that they would let the phone ring for a while when a customer who distributed the sources had a problem.
Since their customers are using it for safety- and mission-critical software, noone would ever dare to cross them.
I mean, I agree, but isn't it a moot point? Isn't part of the reason you get all your contributors to sign CLAs so that you can put whatever license on your product you need to close a deal?
This is a naive view. Most companies try to limit or restrict the scope of what they must open source. With the GPL, depending on the version, how the code interacts with your project (static linking, dynamic linking, IPC, etc.) you may or may not have to release your code under the same license.
In particular Google's strong anti-AGPL stance has caused a chilling effect for adoption of AGPL projects at companies, hence why commercial licensing is frequently an option.
Finally, you can believe whatever you want, but ultimately it's the lawyers who decide, and your ass (assuming you're an IC) isn't the one on the line when shit hits the fan. Lawyers don't like GPLv3 and they don't like AGPL.
This is a post about running an open source company. In other words, you have to worry about your prospective customer's lawyers objecting to your product just because it is in part AGPL. You could just upsell with a different license; but from a business model perspective that makes the distribution channel of open source irrelevant.
What you are describing sounds more like a problem with sales, or marketing, than with legal. I don't think your legal department would prefer you to use BSD instead of AGPL.
There's likely nothing your sales or marketing team can when Google, and companies that cargo cult Google's legal, decide not to allow your product gain a foot in the door because of its license.
> The GPL is one of the cleanest licenses out there. Use the code in your project, give the source of your project to anyone you give the binaries to, job done.
I'd encourage you to read the GOLs and some of the legal commentary about them, if you haven't already. "Clean" is a value judgment, but as one who routinely handles both open source and commercial license agreements, I'd never refer to GPL as particularly "clean", and I doubt I have any colleagues who would.
The summary you give is not complete, overall. And it omits a lot of details that can prove important.
All we would need to do is rename the GPL to something else. That’s it. It really is that dumb. Just don’t tell anyone it’s a renamed GPL and maybe rearrange some wording.
We abandoned GPL for the BSL for this reason. The GPL is in many ways better but there are a ton of companies that won’t touch it or anything that comes near it... though they tend to grandfather in Linux.
I suppose it would depend on what one takes “open source project” to actually mean. You’d have to reject contributions from your community if they weren’t interested in giving them back upstream in a way compatible with your commercial license, and you’d probably be forever open to claims you violated some individual’s GPL contribution of a feature you may or may not have seen if something similar makes its way into your commercial project.
There are obviously lots of open source projects that find a way to navigate this world via dual licensing, and I’m neither a lawyer nor an expert on the subject, but those projects don’t really fit my own personal definition of “open” (as I derive it from the Unix philosophy instead of the GNU is Not Unix philosophy).
Dual licensing requires copyright assignment, largely preventing a community of contributors forming around the software (nobody wants to provide free labor to somebody else's get rich quick scheme), which is one of the big advantages of OSS in the first place.
In a way, dual licensing is essentially proprietary software, with a for free version that you hope will satisfy as few users as possible. A bit like the old shareware days.
Dual licensing also creates the perverse incentive of trying to scare your potential customers how horrible copyleft licensing is. So if you motivation is to increase the amount of FOSS through copyleft licensing, you're shooting yourself in the foot.
The perverse incentive of scaring people about the license only really works if the license is hard to parse or the sales team is really unethical. Unfortunately that has proved the case when firms choose a GPL and hire the wrong people.
If the license is straightforward enough, or comes from a neutral organization that publishes trustworthy lay summaries, as Creative Commons does, there's less room for sales incentives to run wild. The usual conflicts of interest in other business models, like holding back code or doc, making the software hard to use, and so on break down, because there's just one product. Customers pay to use the product.
cutemonster is correct. Dual licensing can and does happen without copyright assignments. It can even happen without contributor license agreements, if contributors make their work available under permissive terms in addition to the required copyleft terms.
GPL variants are more defensible but make open-source-as-a-marketing-tool way less effective. Developers know they can't use GPL software for work, and OSS companies need them to use tooling at work so they become leads.
He obviously means software ( libraries, tools ) which become part of your product. Not tools to build your product ( compilers, bug trackers, operating systems and the like ).
I'm working on an open-core product [0] right now, and an aspect I find missing in the article is that you can generate trust by showing how the internals work. This is especially beneficial if you work in an area like us. Of course, you need to make sure your license is set right and you need to think about where new features go, but it's a valid use-case.
Lovely landing page – there seems to be a sudden abundance of google analytics alternatives by small teams and indie hackers in the last year, any insight into why that is?
They did get more annoying lately though. Nowadays you have to wrestle through a horrible click game of dark patterns to not accidentally accept anything. Things like putting a small "Accept all" in the top right corner where you expect a close button or hiding the settings with at 12px in light grey.
> giving your product away for free is just as bad a business strategy as it sounds.
Has anyone, anywhere, argued the opposite? Because this seems like beating a strawman.
> Copyleft licenses such as AGPL force adopters to publish changes back to the source repo or offer them via network protocols.
The AGPL specifically does this, other copyleft licenses (like the GPL do not for “adopters”, only for people redistributing executables)
> This is meant to deter third parties from offering competing hosted services
No, copyleft in general is about prohibiting downstream proprietary distributions (and also, in the AGPL case, proprietary hosted versions) it is not supposed to prevent competing software/services.
> Instead, why not do what traditional businesses are doing, sell a product, and simply charge for the value it provides?
That's what “open source” companies do. The thing they sell is the product (whether it's support and consultancy, the value-add of the enterprise version of an open-core product, or hosted SaaS.)
The problem with “prop-tech” is often, as this piece itself points out, but entirely fails to address in recommending “prop-tech”: “the world of FOSS is full of strong and established offerings with a zero-dollar price tag.” And making your tech proprietary doesn't stop you from having to compete with the open-source competition.
When I was a broke college student, I cared a lot more about open source. Now that I have more resources, more experience, less time, and a personal interest in getting paid for software, I have stronger feelings about the interface.
Whether or not the box is black isn't really on my mind these days. I rarely have the skill or time to fix them when they break, or don't do what I want. I'd rather the box be easier to swap out or build around when requirements change. I'd rather have a richer ecosystem of blacker boxes, than a bunch of broken transparent boxes laying around.
A problem, of course, is that businesses don't want their boxes to be interchangable. Transparent boxes do tend to help.
For standalone software like databases or CLI tools AGPL isn't an issue, if legal departments would object against using such software those companies couldn't run Linux/GNU tools either. For libraries it's a different story of course. The other points are mostly true, I think if open-source is just a marketing strategy to you it's not worth it.
No, Elastic was licensed as Apache 2.0, which is much more liberal than AGPL. Also, they changed the license because a large company basically resold their solution withtout granting them a share of the revenue. OPs point about the (A)GPL was that it would hinder the adoption of OS software by companies.
> If you are running a commercial business on top of your open-source solution, it is paramount that you own the product's full IP.
Why is it "paramount"? While I have seen that behaviour, there are also several commercial businesses in my corner of the 'net based on open source solutions that don't make an effort to own the full IP and they are doing OK.
I think there's some unstated assumption here, anyone care to explain?
I think the implicit assumption is that you are doing the majority of the development, and you want to make money from licensing, and you want to be the only one making money from your product. So basically, the plan is to make money exactly like a proprietary software company would make money, except that you make the product open source for marketing.
I never understood why people thought that was a good idea. People like Open Source because they don't want to depend on a single vendor, and yet these companies want to be the only vendor of their Open Source solution. It doesn't make sense.
Control. If you control your IP, then the company is more valuable because there are more options (whether selling the company, relicensing/dual licensing or something else).
It's similar to developers assigning copyright when they make a contribution to an open source project--keeps control in one place, makes it easier to make long term decisions about the project (for the project's maintainers).
> If you are running a commercial business on top of your open-source solution, it is paramount that you own the product's full IP.
That sentence is even completely absurd. Suppose you ship a product/deploy a service that includes, among other things, Linux and PostgreSQL. Is it paramount that you own the full IP of Linux and PostgreSQL?...
What is paramount is that risk needs to be reduced, mitigated, transferred, avoided, or even sometimes accepted. Risk must be managed. That is why some companies want to own the OS project, and others want to pay as employees the prime contributors.
The main business proposition of open source is that you can create a larger market. This is hinted at in the article by saying "easy distribution", but it isn't really discussed.
The goal of open source is to create a market for the product that is 10x or 100x bigger than a proprietary product would have. You won't ever capture 100% of the market. But if you capture a fraction of that larger market you still come out far ahead of the proprietary product in which you capture close to 100% of the market.
Probably there are some products for which the market is finite or open source won't make it bigger, and for those the scarcity based viewpoint of this article makes sense.
It is also worth noting that Red Hat has shown that open source doesn't require distributing your product for free.
Another note is that many companies today take an open core approach to try to gain some of the advantages of open source without all of the downsides.
> The goal of open source is to create a market for the product that is 10x or 100x bigger than a proprietary product would have. You won't ever capture 100% of the market. But if you capture a fraction of that larger market you still come out far ahead of the proprietary product in which you capture close to 100% of the market.
The point of the post is that even if you capture a huge portion of the market, monetising it is very difficult, and profitability potential is the main reason you invest.
While open source databases struggle to be profitable, MS SQL Server and Oracle are both very profitable. Why? Strong product-market fit - these products are heavily targeted at enterprise who are willing to pay money, and there is no 'free roll your own' solution to avoid paying the licence fees.
Many successful databases are open source. MongoDB (at least before the recent license change!) is a good example of a huge company where I doubt anyone would have made a noticeable business out of a proprietary version of the product.
I actually think MongoDB highlights the challenges in this article. Namely:
- Tremendously popular database, a huge success by open source standards!
- Their open-source API has been used by Amazon to create their biggest competitor (DocumentDB). Or alternatively any cloud provider will help you spin up pre-built images with MongoDB installed (no path to monetisation).
- Have not (yet) been able to turn it into a profitable business, with large (and widening) losses recorded the last 3 years (at what appears to be the height of it's popularity).
- To create revenue they need a huge sales team (I've heard the pitch and seen the cost!). They not only have to bid against competing technologies, but also their own free tier!
- This resulting in... Mongo having to change their licence to create a profitable business.
23.15B market cap though - but they have had to ultimately change their licensing plan to find a profitable way forwards.
The sad thing is so few companies really invest in participating in FOSS software. That’s one reason these “open source companies” came to exist. It really should be a community of contributors, backed by a diverse set of companies. At best, employers will allow contributing bug fixes or small patches. Not the in depth dedication these projects really need.
We’re heading towards a handful of giant companies that can afford to pay large teams to build and maintain FOSS, and hence own developer mindshare, for their own selfish reasons.
> We’re heading towards a handful of giant companies that can afford to pay large teams to build and maintain FOSS, and hence own developer mindshare, for their own selfish reasons.
"heading towards"? I'd say we're there.
And it's been a while coming. I wrongly discounted Stallman. Like many, I found the GPL onerous and annoying to deal with as a developer. But we've paid a high price.
I don't think it's really a price - including corporate projects was always the point of "open source" as opposed to "free software". The split is generally understood to have started when part of the community embraced Netscape's open source browser.
To be clear, I don't typically have any issue with corporate use of my code. That's why I chose MIT and other permissive licenses.
That said, what I had in mind when I licensed my code were start-ups and small+medium businesses. It didn't occur to me that behemoths like Amazon might stripmine open source and fail to contribute back. Even though the permissive licenses permit this kind of behavior, it seems exploitative.
Additionally, as the GP mentions, we're at the point where it's primarily massive tech companies who can afford to develop ambitious open source projects.
Far more importantly, it also failed to occur to me that tech companies would become the new railroads -- massive monopolies that dominate society, the economy, culture, politics, etc. When I started in open source, Microsoft had recently been slapped down by antitrust regulators, there was no social media, and the tech ecosystem - particularly the open source ecosystem - was much, much different. It was more democratic, egalitarian, and balanced.
Granted, as far as I know, it has been small and medium sized companies that have used my own open source code. I'm not aware of any of the Big 5 using it (my most popular libaries have been superceded by newer/better projects anyway).
Nonetheless, I would have been more wary of philosophically supporting permissive licenses had I fully realized the implications.
> Far more importantly, it also failed to occur to me that tech companies would become the new railroads -- massive monopolies that dominate society, the economy, culture, politics, etc.
These are social and political problems, you can't fix them by sitting in your basement writing copyleft code anymore than you fix them by writing permissively licensed code.
It's not like the FAANG's of the world would go, oh no we really want to use this copyleft library, oh well, let's release all our code under a copyleft license as well. They'd just use some other code with a more permissive license, or then they would reimplement the needed functionality themselves.
> These are social and political problems, you can't fix them by sitting in your basement writing copyleft code anymore than you fix them by writing permissively licensed code.
This is true. But just because you can't fix something by yourself, doesn't mean that you should contribute to, or participate in, the problem.
Have a look at PostgreSQL, SQLite, LLVM, Clang ... all these are projects that get a lot of support from proprietary software vendors. I'm pretty sure that the companies contribute to these projects because of the liberal license, not in spite of it.
I think a big fallacy with regard to the GPL is that companies don't contribute to open source because the license forces them to. Companies contribute to open source because it makes more sense to collaborate on some projects rather than each work on their own.
Licenses like GPL etc. are great for ensuring their user's freedom; but they don't really incentivise companies to contribute to them at all. Instead, they prevent some companies from using them. So counter-intuitively project with liberal licenses end up with much more contributions than projects with strong licenses...
Counter argument: the Linux kernel. (Since we are mentioning open source projects with the highest profile on earth, I don't think neither of the examples are fair)
I agree, the Linux kernel is one of the examples where the GPL really works, because a lot of companies really need it for their products, and they need to distribute it, and there are very few alternatives.
QNX, WxWorks, Windows Embedded, Windows CE (RIP), NetBSD, FreeBSD, OpenBSD, Solaris, OS/2 Warp, Plan 9, Inferno, RiscOS, ReactOS...
It's a shame what happened to BeOS - I wonder what would have happened if BeOS was open-sourced when the expected Apple buyout fizzled and the company went-under instead of it simply disappearing - that would have put a huge damper on, say, Ubuntu and other Linux desktop projects because the whole selling-point of them being "the only usable desktop alternative to Windows on non-Apple hardware!" wouldn't be true.
Well, there are also many who want to contribute code back to GPL and LGPL projects, but can't because the companies that control these projects want them to sign contributor agreements with unacceptable terms.
Probably a big reason why Linux is successful in attracting contributions; the distributed copyright means that there's nobody more equal among equals that has the sole right to monetize it however they see fit. Everybody gets the same deal.
> We’re heading towards a handful of giant companies that can afford to pay large teams to build and maintain FOSS
“Heading toward”?
I remember people talking about that as the existing state...in the 1990s. If anything, we’re heading away from that, with a much wider array of companies actively investing either cash or employee time in open source.
What's actually changed, IMO, is that it's a lot harder to get traction with commercial software than it used to be, so even forms that have no open-source-compatible business model in mind launch as open source, meaning that they have to figure out a unexpected business model or pivot out of open source at some point.
And, for PR reasons, they want someone else to blame for the pivot out of open source, so we get all the AWS-blaming and “the market has become radically less favorable to open source” stories, neither of which is grounded in reality.
> We’re heading towards a handful of giant companies that can afford to pay large teams to build and maintain FOSS, and hence own developer mindshare, for their own selfish reasons.
Can you name some examples of such "selfish reasons" connected to some OSS?
I can think of Google taking over the internet with Chrome, but MS did the same with IE, so that has nothing to do with OSS. It's more that Chrome happens to have source code available.
A community of contributors backed by a diverse set of companies has to be seeded by an MVP. Where would that MVP come from? For complex software it's either an internal project spun out from a unicorn or a labor of love from a gentleman hacker. These are pretty limited routes; it's much easier to start an open source startup with the caveat that the community will probably never emerge later because the startup will be moving too fast for outsiders to jump in. Now startups feel they have to use "innovative" licenses to be successful but the powers that be are deriding these licenses as "fake open source". We're setting ourselves up for an OSS winter where sophisticated developer tools or installable infrastructure software just won't be created.
The term "open source" has a well-known definition, and the "novel licenses", like Mongo's, do not match it. They are more like the "source available" licenses, as practiced by MS in 1990s, and by IBM, much earlier.
There are several definitions for Open Source that come from foundations and other groups, which push their own beliefs and standards. Opensource.org, for example, pushes their own definition that excludes all those who do not use the perferred license!
However, dictionaries that mirror collective understandings, as used by millions of people, tell a different story. According to OxfordLanguages, Open Source denotes, "software for which the original source code is made freely available and may be redistributed and modified." If you look at Dictionary.com. you will see that Open-Source is defined as "pertaining to or denoting a product or system whose origins, formula, design, etc., are freely accessible to the public."
So, no, open source doesn't have a single definition, except to groups who are pushing their own definitions.
> There are several definitions for Open Source that come from foundations and other groups
...and all of them do so because they really want to pretend that their software is "open source" without actually putting it under an open source license. Like, in the abstract I agree that it's weird having the OSI (or any one group) define "open source", but in this case I'm okay with it because every single "alternative definition" turns out to be someone just trying to pretend that their source-available software is "open source" for marketing reasons when it isn't.
For software the overwhelming consensus is the OSI definition (extremely close to the FSF def, extremely close to what is applied as criteria by various distro, etc.)
You may want to write "Open Source" to be sure it is not confused with something else, but even "open source" should be parsed as that definition by default. This is not even pushing a political agenda or a judgment of value. Just "everybody" calls that like that, so there is no point in overloading the term for other approaches. Using another term for other things is perfectly fine and there are plenty of them possible with a positive tone, in case they bring positive things to the table.
Interesting post - I'm coming from a COSS database perspective (TerminusDB), so talking my book. The point of COSS imho is to create broad value and monetize a percentage of that value. A few points:
1. In database land, there is no other available venture scale business model. If you look at the top 10 databases on DB-Engines (https://db-engines.com/en/ranking), the most recent proprietary db is SQL Server which launched in 1989. Yes, databases are sticky (we see you Oracle), but unless you believe there are no further investable plays in databases, then open source is the only way to invest. (Snowflake is an exception to the rule, but in this case, it is just that).
2. Useful recognition that offering hosted versions of the product is a viable business model that can deliver SaaS type margins. Mongo's Atlas is the obvious leader here - a huge and fast-growing revenue stream. I would also mention the companies that have built successful SaaS/marketplace offering on OSS. Informative list of all the COSS companies above 100m in revenue:
https://docs.google.com/spreadsheets/d/17nKMpi_Dh5slCqzLSFBo...
3. I find this conclusion difficult to swallow: 'The ratio of failed OS businesses to successful ones is worse than in prop-tech; revenue kicks in much later'. I can't find any evidence for the first point - it needs substantiation. For revenue, lots of the most successful venture backed businesses (facebook) have revenue kicking in later so they can maximise growth potential – that’s seems ideal for venture!
If you give away your secret sauce recipe, no one will pay for your recipe (though some will donate), but people will still buy your fried chicken.
I think open source works best when it isn't your product, but your capability.
An e-commerce site open sources it's product inventory management software. If I spin up a docker container with your software, I will be closer to building an e-commerce business, but I won't have one without a lot more work.
More examples:
A blogger makes her own open source blogging software.
A 3D games developer - open sources their AI engine or 3D engine.
You get community adoption, goodwill & hopefully contribution, ability to hire employees that already know your software. Possibly it identifies new markets to pursue.
I'm working on an "open core" knowledge base [0] right now and have been for the last few years. We're definitely leaving money on the table by offering an easy docker install option and I'd largely agree with the points made in the article but it's hard to understate the value of the distribution that open source brings, particularly for product-lead teams that aren't so hot at marketing.
If the software fits into the 'infrastructure' bucket then open source software seems like the best approach for selling into engineering orgs. This is likely due to the longer setup/eval time for infrastructure type software (Elastic being an example) and the higher price tag (justified or not).
Contrast this with 'tools' (such as IDEs) which tend to not be FOSS and are a low enough dollar and time cost that they can easily be purchased or even trialed for a month.
My very brief review of your product site leads me to think that a good approach for your company would be to offer small-team licenses for free with the hopes that they will adopt and become champions for their enterprise to buy a site license.
I have debated over this for years while building a self-hosted analytics platform (userTrack.net) and I have always reached the same conclusions: if I open-source it, not only I affect its sustainability but also the business goals would become complete opposite to the product goals (eg. as you mentioned, making it easier to install improves product but affects the business).
I now think that the best way to go is "paid-source": make the product as good as you can, make it as easy to install as possible and the customers who want to use it can pay a reasonable amount for it. I recently changed the pricing model, as the one-time payment was not sustainable, so the product has now an upfront cost + optional yearly maintenance fee (you can use it forever, but if you want new features and support you should support it).
I have used many amazing open-source projects that got abandoned over the years, I would have gladly paid for many of them if that meant sustaining the future development of the project. And, before anyone mentions it, donations are not the answer.
> I now think that the best way to go is "paid-source"
Is that just source-available software[0] (sometimes also called shared-source)? I ask not because I object, but because it's easier if we all use the same words for the same things.
I think the pricing you mentioned is also part of the issue.
Companies that are open source but paid something have a crazy high price for that something. I know they are going after the Enterprise market but they still price themselves out.
I also found out the same thing, but it's mostly related to paid software in general. It's a lot easier to sell expensive software. The lower your set the price the more "cheap" customers you will have, which results in more support tickets, maintenance, marketing needed, etc. The customers who want lower prices are also those who would be more critical of your product, as the price they paid for it means more to them compared to someone who doesn't mind spending $50/mo for a basic service.
I am still trying to find a way to accommodate for both enterprise customers and low-budget customers. I tried doing this using some simple licensing terms (eg. get this if you are a solo dev, get this if you are large business), but people will just buy the cheapest option that has the features they want. This makes it hard to offer the same features at different prices and might lead to either downgrading the basic product or adding some buzzword features for the enterprise product.
We live in a fundamentally capitalist society. Venture capital, entrepreneurship, the startup culture, etc, are all fundamentally profit driven.
Open source is a movement which is ultimately driven by ideology. Even open source advocates need to put food on the table.
While there are business models that can support open source projects, these will always be the exception when humans are involved. There is always an entrepreneur willing to ignore ideologies (and sometimes laws and/or morals) to make a buck.
I believe that there is a need and important place for open source and free software, but the fact is that this needs to be funded by philanthropy and dedication to ideology, at the exception of profit in most (almost every) case.
In this context, I think the article is correct. Depending on a majority of people for good behavior, good will and altuism is naive.
I always see FOSS to get super fast traction (Elastic) and userbase - hey, because its free - and then later the same people complain that their lunch got stolen. Elastic could have grown (or not) organically with closed source software but that's hard without community help and being able to hire top developers to build your product. FOSS seems like a great way to market/PR your name out there because free is popular. I compare this, perhaps not perfect analogy, to Uber. Offer cheap rides, gain popularity while burning investor cash and wipe out existing industry/competition. Then, investors complain they can't monetize it. No shit! In the case of Elastic, the rest of the world benefited from it atleast.
Algolia is doing just fine. Open source contributions by volunteers are typically very small compared to the work done by the corporations behind the open source projects. At least for the vast majority of projects.
> Elastic could have grown (or not) organically with closed source software but that's hard without community help and being able to hire top developers to build your product.
Yea I don't know. If I saw claims made at [1] by a startup that made proprietary SW, I'd think "yea right".
The fundamental difference between buying or not buying OSS products is that if you’re buying OSS, at least there’s a potential your product will be fixed and live on through the community if your original vendor disappears. The only thing the vendor can really deny you is the service. In the case of closed source, you can not only be denied the service offered, you also end up with a useless, unpatchable black box if the vendor disappears. In my mind, this makes OSS more resilient, in more ways than one.
This doesn’t contradict the point made: of course from an investors point of view I’m sure closed source is often more profitable. But from a consumer perspective, I doubt the same benefits are true.
I think "Commoditize your complement" comes into play here. Yeah, if your core offering is free, you're going to be fighting to make money. But you can get huge value if you open source otherwise proprietary parts of your stack.
We've done this at my current company. We have a closed source, commercial offering. But our client libraries, our documentation, sample deployment scripts, and other supporting libraries are open source.
The essential problem is gatekeeping. Any commercial business uses money as a means of getkeeping what it uses (buys). This works well to reduce and so control the purchases of chairs, desks, external consultancies and proprietary software.
We all know the horrors and pain of trying to raise a PO to get a license for a really useful / necessary piece of software.
I fundamentally guarantee that if React needed a 0.01 cent license for every internal and external installation it would have four users globally.
FOSS short circuits this gatekeeping function - and it is only popular because it is hard to keep track of and easy to load up npm and grab 900 libraries.
Imagine there was a FOSS plug-in everyone used, call it mothership that on every install (and indeed invocation) it routed back a simple packet to say "i am alive and used on this machine in this domain in this company" (let's call it a common piece of config like git email config)
Immediately every legal department would soil their pants and internal software development would grind to a halt - or FOSS developers would be able to charge decent value for their work - or some ecosystem of foss-aggregators would spring up providing licensed, limited installs and support and legal protection.
Personally I am in favour of the last option - for each product a series of (local) companies could club together and provide support for each other's code bases, (if you use my react-foobar I will license company X to indemnify you - sort of approved vendors schemes)
I struggle to see many better ways
Edit:
Andressen was slightly wrong - FOSS is going to eat the world. And as such it's production will be a public good - paid for out of taxpayer money and directed as such - either as a utility through regulation or as science through funding only the brightest.
And the big question remains - if we could see the whole stack easily, and if it all was priced at the minimum wage for every developer in the stack how much would it be?
Would anyone pay it? Or worse, would they realise they did not need all those in-house developers anymore as they are all running "libBusiness.py" and it all kind of fits together.
Or as is more likely - will we see lots of companies like redhat, that produce a "bank/airline/retailer in a box" - and the internal teams are focused on that 5% of USP?
I am not sure I understand that world - but I also do not think I currently could select correctly all the FOSS packages needed to build that stack that can just run a business.
Trust me it does not look like the openERP style stacks. at all.
McDonalds has a operating manual - it details exactly how to cook the burgers, how to operate the trash compactor, insurances etc etc. It is how to run a fast food restaurant and it is I am sure proprietary information.
But if they published it under creative commons they would probably not lose a fortune.
Firstly there is already completion in the market - plenty of small one person late night takeaways exist - all of whom have some level of floor quality enforced by regulation, taught by suppliers or simply being fairly obvious.
The mcdonald's effect comes partly from consistency yes, but also supply chain efficiencies and huge marketing costs.
So do we ask, is a Software house that chooses to release FOSS or not actually Mcdonaldds - or are they the department inside mcdonald's that just writes policy operations manuals?
A small team of developers will find it hard to just write code and make a fortune / business / change the world.
two of those are either need a purchase order upfront, or in 30 days (which is waaay shorter than the time it takes to actually raise a PO!)
and ... who gives free as in beer but not open source software? I mean winzip is all that comes to mind?
But it does not change the essence - only FOSS gives you software "forever" whilst jumping over the otherwise immutable accounting department t - it is probably the greatest hack no one sees.
If you got a lot of money you should give back to what propelled you there. You may call it investing with a loss but that'd be just the right thing to do. Everything is not about the money, making money and more money.
Cryptocurrency issuing entertainment companies have monetized this pretty well.
They don't have recurring revenue but they don't need venture capital investors either so it doesn't matter.
Open source, campaign, sell, book revenue from sell. If market has formed, book more revenue in the future from selling into liquidity. If market hasn’t formed, hire some entertainers to babysit a chat room and move on.
Some parent companies are monetizing repeat issuances, those are worth investing into because all the shareholders make money. The companies dont need capital and would only offer shares to provide some exposure to the market. I’m not aware of any broadly offered dealflow of a token issuing factory, which is probably why people don't understand or notice this business model.
Let me say this again. Open source is not, in and of itself, a business model.
People conflate having an open source project with having an open source business. Open source businesses can be done, of course, but open source in and of itself won't give you that business. It is not a business model therefore but rather a distribution/governance model.
The key is that being open source should not be a core value proposition in most cases (unless you're Red Hat) but rather it should be ancillary, merely incidental to your core business. For example, say you ran a task management app company. The core value driver is not whether it's open source or not (to most people). People don't buy based on whether the product itself is open source, they buy because it solves their problem, of task management in this case.
Labeling oneself as "open source" is a classic engineering trap because again, people don't care about that, if a closed source solution can solve their problem more than an open source one, they'll use the former.
By the way, I happen to run an open source task management app (a todo list + calendar hybrid basically, https://getartemis.app) but nowhere do I advertise it as open source, even though it is, because I know my core customers, non-engineers, don't care about that and just want to schedule their days better.
---
I am reminded of OpenHunt, an open source version of Product Hunt that was made. The only value proposition was that it was open source, but it shut down precisely because users didn't care about that, they cared whether it had enough cool products (as a consumer) and enough traffic (as a submitter).
> OpenHunt tried solving a problem for the content makers without providing any additional benefit to the content consumers.
> It's a nice, heart-warming mission. But in the end of the day, content is king, that's what consumers want.
> There have been many examples of people rallying around a "free and open" version of a service. They fail to realize that the end consumer barely cares. Look at voat (Reddit), app.net (Twitter), Diaspora (Facebook), even ycreject.com (Y Combinator) tried to be a thing for a while.
> If someone is able to make it "free and open" while also making it a better experience than the alternative, then it'll be a big success. But so far everyone gets that wrong.
post points out that there is variation among what constitutes open source companies / potential for profitability, and yet self-contradictorily concludes a simplified blanket rule on open source being a 'better strategy' and uses this simplified rule to conclude a similarly simplified blanket rule for investing in them..
I was at a conference a few years ago and the conversation around the lunch table was the chilling effects of the public cloud's operational expertise on venture funded open source businesses (what we saw with Elastic). Nothing I've seen since that conference has changed my mind.
As a developer, I love using open source software. Easy to review/bug fix, less risky because even if the company fails, it'll be around, and usually very low cost and easy to get started with.
As a prospective business owner, I think it stinks.
Another view: as a consumer of software, I love free stuff. As a producer of software, I like to get paid.
The ratio of failed OS businesses to successful ones is worse than in prop-tech; revenue kicks in much later, business model pivots are hampered by community resistance, and licensing issues leave OS businesses vulnerable throughout their lifetime. Instead, why not do what traditional businesses are doing, sell a product, and simply charge for the value it provides?
Why not? Because when your project gets so big that lots of people rely on it, it would be better for the world to remove the private property restrictions, allowing fixes and innovations to come from anywhere and be distributed as widely as possible.
The question should be asked the other way: why not make it open source?
You see, all the objections here come from a capitalist mindset, which is based on private ownership and competition. “I built it so I own it.” That is how we ended up with a feudal society online, where conservatives are complaining they’re being silenced, where bulk collection of everyone’s info is possible in one place, where rents are being extracted from the ecosystem.
The alternative is collaboration with no private ownership of the platform. (Note that this is NOT socialism.)
Linus Torvalds launched Linux, Tim Berners launched the Web, Vitalik launched Ethereum, but they don’t “own it”. And it is precisely because there is no private ownership of the whole platform, that they have led to an explosion of wealth and utility for the entire world.
“Business models are incompatible with open source”. Well, sure. Rentseeking behavior gets harder when the platform is permissionless. But what has led to more value for the world:
AOL or the Web?
Internet Explorer or Chromium?
Britannica or Wikipedia?
Windows or Linux, BSD (including MacOS and Android)
Alchemy or Science?
This is a mindset issue in our culture. In current society the vast majority of people are living paycheck to paycheck, so they would love to escape this precarious situation. They take jobs because they pay, not so much because they like the project. When they do get financially independent, they keep going and going, trying to make more money and become a capitalist. And we all support these few people who “escaped and made it” by ensuring that their intellectual private property rights are enforced and their business models are enabled in our society (RIAA, MPAA, etc). In a society where everyone has UBI, people would work 20% of their time on stuff they like - science, open source, religion, hobbies, raising children etc. And projects would attract people because they are awesome and useful to many people, and because they can contribute something that others need.
Each small contribution to the gift economy costs very little, and there is no limit on the remixing and reusing of effective solutions. Private property is all about adding more friction to this permissionless model and excluding the rest of the world from using a resource and building on others’ discoveries. When Isaac Newton said “if I saw further, it’s because I stood on the shoulders of giants” he didn’t rent those shoulders. He wasn’t sued an a SWAT team didn’t break down his door. Isaac Newton wasn’t in pharma.
If I sound like some kind of communist, liste to the venture CAPITALISTS who fund the future and see where things are doing. For example read https://worldaftercapital.org by venture capitalist Albert Wenger general partner of Union Square Ventures to understand where we are headed as a society. And read their latest 3.0 thesis.
I mean you are essentially selling yourself short with FOSS unless you were in a very saturated industry and selling support was the bulk of your revenue.
Wow. It looks like this man has no idea what the free software spirit is. I mean, he definitely hasn't, as he's talking about "exit value“. Still, he's writing that with HTML, using SSL and TCP, on a PHP website probably running on Nginx or Apache on a Linux server with a postgresql database... Opensource is what allows him to talk to people or even create his business he is so proud of that he gives advices to everybody, yet he encourages everyone to not join the game. If open source has some problem, it's certainly because of people like him.
Do engineers working full time in a FOSS core team need someone to pay their salaries? Nothing the author said was against FOSS on a volunteer basis. What he said was it’s a challenging business model, unless you can reach escape velocity, and the window for that success is much narrower than simply selling a commercial product from day one.
There are real differences between FOSS that provides foundational capability (network layer, drivers, OS) and application layer projects. Firstly, they largely were built at a different time, secondly they were pure FOSS and not backed by a founding company that relied on it to keep core development going (Redhat’s business backed into Linux, so they benefitted from an existing core).
TLDR a founder who struggled to establish a business based on FOSS is saying don’t do it as a business because it’s harder than you think and can feel like a thankless endeavor.
When I worked at reddit, we open sourced our code. It didn't gain us much. We got a few contributions, and we also got a few competitors. The biggest benefit was being able to show people the code when they suspected shenanigans.
At Netflix our ethos was "if it's involved in infrastructure, open source it, if it's related to movies, keep it closed." That actually worked pretty well. But the open source code was difficult to use unless you invested in the entire ecosystem. The main benefit was recruiting.
For my own company, I try to support open source by donating to the people who make it.