Amazon Has Gone from Neutral Platform to Cutthroat Competitor: Open Source Devs
onezero.medium.com
onezero.medium.com
When its convenient for these companies, they're more than happy to release expensive, closed source addons to their "open source core". When its convenient for them, they'll point a finger at Amazon and say "look at how evil they are, they took all the hard work we put into this and want to sell it" (simultaneously, the hundreds of unpaid community contributors in the audience look perturbed, raise their hands to complain, and are immediately shushed, "just ignore them")
"Open Core" companies like Docker, Mongo, or Elastic want to have it both ways; they want the moral high-ground that comes with accepting free contributions from the community, alongside executive power when it comes to deciding what will and wont be open source and who can and cannot use it for what purposes. You shouldn't have it both ways. That doesn't mean you can't build a successful "open core" company (all three of these companies are successful), but lets drop the act of granting them moral superiority. The only reason they exist is because of the exact same licenses and policies which enabled Amazon to do this.
[1] https://github.com/mongodb/mongo/tree/master/src/third_party...
When will the corporate shilling on HN end?
The only difference -- as you pointed out -- is size.
"Might makes right" is just as untrue as its opposite.
The core piece of ElasticSearch's technology is Lucene, an open soure project. Elastic the company uses this open source work to become a 5 billion dollar company.
The core piece of AWS's ES service is ElasticSearch, an open source project. Amazon the company uses this open source work to further enrich themselves.
Amazon is bad because they're a 1 trillion dollar company. Elastic is OK because they're a 5 billion dollar company.
> Yes, the whole argument is based on anti trust, as the article explains.
Building blocks and libraries offering convenience are not as valuable on its own than solutions that offer some quite immediate value at a reasonably high level. Even if from an abstract view both still require developers as users the applicability is quite different. Amazon cannot 'host' lucene as a service per se.
Services vs Libraries
Your argument makes no sense. Even if we accept the thesis that building block components are up for grabs because they cannot be immediately used as valuable services... Well, neither can services. You need to design a system architecture, figure out how and if a component plays a role, pick a component, config the component, setup a deployment ans scaling strategy, config and integrate metrics and health checks, ingest data, monitor the service, etc etc etc.
And if we actually want to make an objective comparison between libraries and services... Well, in the end they are all components of a larger system.
So, obviously there is no objective distinction. There's only intellectual property and companies being hypocrite wanting to have their cake and eating it too. They can't get it both ways.
Their mongodb clone isn't. It'd be kind of cool if it was, but there's substantial evidence its based on the Aurora Storage engine. Through the docs and various talks at Re:Invent, its clear that the ASE is insanely coupled to AWS, with tight integration to S3 and deep availability-zone awareness. It would be useless beyond an academic exercise to someone who isn't on AWS. It would be cool to see it all open sourced, but not really valuable.
AWS has a pretty strong history of open source in the cloud space, such that it feels strange to me that people still think they're an "enemy" or something. Not as strong as Google or Microsoft, but they're certainly not a New Oracle or anything. You should check out their Open Source Blog [1], and notice that new posts are made literally every two days.
I had to laugh at this one. Isn't this what many open source projects have been doing forever? Open Office, Chrome, Apache, etc. I could go on. Many open source applications were designed to compete with a paid/proprietary counterpart.
"It is clear that AWS is using its market power to be anti-competitive"
So offering a free product and adding value to it/offering it to your customers for free is now "anti-competitive"? Isn't this only beneficial to customers? We should want more companies to do this.
This article has some good info on this fight: https://www.datanami.com/2019/03/12/search-war-unfolding-for...
It sounds like Amazon wants to add new features to Elasticsearch, which just so happen to be proprietary features that requires a license from Elasticsearch.
This is always the risk when basing your entire business on an open source project. A much larger company with more resources could build the same features and release them under the same open source license, for free...which is what's going to happen here...and it's actually exactly the spirit of the open source community.
This is why I would never create a business based on open source software and attempt to have a dual licensing model. In the end, the only real way to make money is by having better support (because any big company can come along and eat your lunch)...which is much less scalable and requires many more employees.
No, you misunderstand completely. There is free software and there is open source. Open source does not compete with paid/proprietary based on price.
You can sell open source for thousands of dollars. Redhat does it every day. But part of being open source is providing source, letting the user decompile/recompile and modify the product you sold them and that they own. If I wanted to strip out analytics services from a browser I can do that in Chromium and Webkit. I can't do that with Chrome or Edge.
> Certainly many contributors to open source projects have been motivated by not wanting to pay for software.
That hasn't been the case for most projects I see. Most developers can afford to pay for proprietary software. Yet they choose to work on opensource, why is that? I can purchase 1 month of Adobe CC with less than 15 minutes of work. Why spend years, months and days developing and debugging an open source image editor?
> most of the users of open source software have been motivated by not wanting to pay for software as opposed to a desire to read or modify the source code.
Most users would pirate the cracked version of the proprietary software (i.e. Photoshop, Windows) rather than use something like GiMP or Ubuntu. If money is the motivation there is a lot less friction in sticking to cracked Windows than opensource ArchLinux.
> Freeware has been around since the beginning of software, open source is a more scalable way of doing it.
Not sure what this means. Freeware can be put on an FTP server or repo and and downloaded quickly. What is the scaling issue?
This is just not the case anymore. I know tons of people who choose to work on open source and I pretty sure none of them did it because of money.
I guess it's might be like that 25-30+ years ago for the people who started GNU and Linux ecosystems when vendor lock-in was big deal and a lot of programming tools were proprietary and expensive include compilers.
And 14 years back when I started to work in web development there was similar situation where a lot of tools and software is proprietary and expensive. Not anymore.
> And many if not most of the users of open source software have been motivated by not wanting to pay for software as opposed to a desire to read or modify the source code.
For initial adoption this is still true, but I believe many people who stick with open source products doing it for entirely different reasons. As others pointed out it's might be not the case in rich 1st world countries, but everywhere else people just go with illegal software instead of open source.
Open office is a good example. Everybody I know uses it as a free alternative for Microsoft Office. How is this not competing on price?
From a developer standpoint, it's great to be able to tinker, check for spyware, etc. But 99% of users will never actually care or look at the source and only even care about it because it's a free version of a proprietary counterpart.
"You can sell open source for thousands of dollars. Redhat does it every day"
You can. But you have no competitive advantage because you are required under the GNU to release your source to everyone (and any other company can sell it for X amount less).
Companies need to either sell support or something else to stay relevant (redhat sells support, certifications, and books).
This is one of the reasons SaaS have become so popular over the last couple of years. You don't need to release your source and could base your entire business on open source software without having to release anything back and lose your competitive advantage.
Perhaps some of us don't actually like Microsoft Office, and prefer things created for public benefit?
The logic behind complement commoditization [1] is to kill your fine-grained competition, when you can recoup the commoditization costs in the larger product. It is something every company above a certain size does. There is no fundamental problem with the practice.
The commoditization practice is problematic when your company is so large it distorts the market. Hence the antitrust perspective of the article.
From a monopoly prevention perspective, Amazon should not be allowed to commoditize complements.
[1] https://www.joelonsoftware.com/2002/06/12/strategy-letter-v/
Step 2: Be shocked when Amazon uses your open licenses exactly how they're allowed to
Step 3: Write angry letter about Amazon monetizing the project you were trying to monetize.
Open source isn't meant to be a get-rich quick scheme for founders to generate cheap buzz off their 'free' project. It's literally designed to be forked and used in whatever way the user/developer sees fit. You can't just unrealistically hope that a large company won't exploit it in whatever way they can.
You choose closed source and you throw away early adopters - people are way, way less likely to try out closed source software from what I've seen. They'll even resent you for it.
You choose open source with a permissive license - you bootstrap, get users, and some larger company comes along and essentially takes it from you.
You choose GPL3 or some less permissive license and no one will ever use your software.
How do you see this working?
Services
Open source is a choice. You get a lot of early adopters, yes, but the project is also completely out of your control. There's nothing stopping anyone else from taking it, putting more resources into it, and stealing your thunder.
If you're not happy with that idea, just don't make your code open source. You can't have your cake and eat it too.
What I don't like is the companies that create some open source project for the hype, then charge for a 'management console' or some marginal add-on.
It's in every way a better product than the hosted AWS ElasticSearch service, which I find to be practically unusable. More importantly, I don't know anywhere that uses AWS ElasticSearch where ElasticSearch is a core component of someone's platform.
The problem Elastic has is that their Cloud isn't remotely as good as self-hosting. Using their Cloud doesn't abstract away the domain knowledge needed to operate ElasticSearch at scale, so anyone who uses it and becomes competent will eventually self-host.
If they can get better at abstracting away the arcane knowledge needed to operate ES, AWS won't be a problem for them.
Gitlab is a good example.
more attractively is the all code for free model, so those who are able can use it for free, and you charge a fee for people who are in need of professional support.
a third one is still all code for free, but you can pay for hosted service instead of self-host the code, time is money for many people these days.
in all cases, you're building a group of followers with OSS, and you sell the hard-to-grasp or time-is-money-for-some-user domain expertise.
how to build your moat as far as competition goes? anyone that is smarter than you could copy all the code and do the same business model, the only thing might be a good license? however these days most OSS license are commercial friendly so I don't know how this works.
On the one hand, I really do understand the needs here, on the other I wish there was a better option.
I wouldn't even mind seeing something like the new MongoDB license + open-core as a non-profit organization that funded the development of several shared projects, where what licensing it does get is directly towards development. It's much different when you are absolutely profit driven.
There are a lot of software problems that are hard, but the results can be shared.
> AGPL meant companies who wanted to modify and run MongoDB as a publicly available service had to open source their software or obtain a commercial license from MongoDB
With the new license:
> The only substantive change is an explicit condition that any organization attempting to exploit MongoDB as a service must open source the software that it uses to offer such service
Aaaaaaaaaaaand from the license text itself https://www.mongodb.com/licensing/server-side-public-license:
> “Service Source Code” means the Corresponding Source for the Program or the modified version, and the Corresponding Source for all programs that you use to make the Program or modified version available as a service, including, without limitation, management software, user interfaces, application program interfaces, automation software, monitoring software, backup software, storage software and hosting software, all such that a user could run an instance of the service using the Service Source Code you make available.
That's true of open source under "permissive" licenses like MIT, BSD, Apache, and now Blue Oak. It was never true of open source under "copyleft" licenses, strong or weak. Others are free to fork and use copyleft software if they release the results likewise.
When companies choose a permissive license for maximum adoption, and then decry competitors for adopting, I haven't much sympathy, assuming they had the resources and wherewithal to make a better choice. But copyleft was also an available option, and copyleft is plenty open source.
I wonder why aren't they standardizing an open source license that prevents hyperscale cloud monetization by someone other than the originators? Don't OSS communities hold the cards here? it's just a change to license.txt, broad OSS adoption and it's a very different type of cloud (picture nuclear mushroom cloud)
I've read quite a bit from OSI stalwarts about why they take the positions they do, but I still don't understand them, and therefore shouldn't try to relate them.
If I had to make one criticism of the OSI mailing list reaction to both my License Zero Reciprocal License, which later became Parity, and also MongoDB's Server Side Public License, it's that takes almost invariably focused on who was proposing the license, and the business model they seemed to have in mind, and ignored the extent to which older licenses had provided substantially the same functionality, as the basis of similar business models, before the software-as-a-service shift. It wasn't clear whether that was relevant criticism or not, since the process wasn't clear to begin with, and notional clarifications from OSI staff and eventually the board only muddied the waters.
When literally anyone can object to a license, or block the approval process, for almost any reason, so long as they can phrase it in vague philosophical terms, then no new ideas in copyleft will ever get through. Even if those new ideas boil down to maintenance work on existing licenses now showing their age.
I missed that you posted this, Kyle. I think that OSI is worse without your participation, and your criticisms hold a lot of weight for me. I hope that the new board, and all OSI members, will take what you wrote to heart, because stewardship does require caretaking, cultivation, and maintenance.
I would love to see more productive discussions than the "how many angels fit on the head of a pin" that we so often see on mailing lists, and I think you're right that the Open Source Definition may not capture what is needed to help those discussions be more productive.
It's really a tragedy of the commons.
Users on average refuse to pay for software/web apps directly so companies have to find other alternative ways to get funding and make money.
People complain about Facebook constantly and how they're selling their data yet these users wouldn't sign up to pay $2-3 per month for Facebook directly.
I'd sign up for ad free FB for $2-$3 a month if it made it easy to set retention periods on my data and opt out of the collection.
So you're offering to pay far less than they make off you right now.
Also that seems way way up since in 2017 they made only 6.18$ in Q4[1]... so who's suddenly paying so much money for our user data?
[1] https://www.cnbc.com/2018/01/31/facebook-earnings-q4-2017-ar...
One day I'm going to build this premium Facebook clone that respects user privacy!
Come on Hacker News, make it so I don't have to!
I cited a number that is annual and is looking at North American revenue which is an indicator of mature market potential.
I block ads on Facebook web and don't have any of their mobile apps installed, so I can't imagine they're making more money from me than if I paid them to not have any ads.
They make money by being able to target ads, not just from simply showing them, so all the data they have about you, and your links to other people, and correlations that can be derived from all that, and so forth, all feed in to that earning potential.
You might not be seeing the resulting adverts, but can you say that about all your direct contacts, and all the people in groups you are linked to, and all the other people who interact with posts & events you comment on or otherwise interact with, and all the people that...?
You are still very much part of their income model by being part of the data graph - you are not breaking the system remotely as much as you think by arranging not to see adverts.
Letting people pay to not see ads would require both initial and on-going development & testing effort so you have to include that in any cost/benefit analysis of the situation too. And that design & development could be pretty complex: by making a legitimate ad skipping option available they risk creating a hacking opportunity that makes it easier for people to skip them without paying. It might also send a bad signal to the advertisers: people with more disposable income are more likely to pay to skip the adverts, because the cost will feel less to them, and people with more disposable income are usually those that the advertisers usually most want to reach.
That advertisers might want to reach me isn't especially relevant, because there's no situation under which that is a possible outcome for fb.
Who the advertisers might want to reach is relevant because FB can currently promise to do whatever is practically possible to block ad blocking. This doesn't amount to much (tricks like the extra divs littered through the word "sponsored" and so forth, to obfuscate the advert related code & markup from blocking scripts, for instance) but is better from the advertisers point of view than them admitting it is going to happen and they've given up stopping it. And again, the official ad blocking method is likely to make unofficial blocking easier - they would have to defend that to their advertising customers.
The other issue that could come up is effectively creating a two tier system, with those that can afford to pay and those that can't. Equality and privacy defenders might be all over that, and however right or wrong people think each side would be there FB are not going to want that on top of all the public scrutiny they are already getting.
Such a system could be easily abused, however, without some kind of bulletproof third-party auditing in place.
If Google offered an ad-free, tracking-free version of all of their services (maps, gmail, search, docs, youtube, etc.) I might even consider using it.
Gsuite?
It's too late for that, I'm afraid. Google and the others have invested billions in their ad products and made back many more billions. It wouldn't make any sense for them to offer any kind of paid opt-out.
And even if you didn't use any Google services, they already have reams of data on you based on your correspondence with anyone with a Gmail account, plus whatever data they can buy off your carrier network and your credit card provider.
Netflix etc do it with premium services etc
I always kept quiet about this, but I've felt that way for a long time about anything that comes from Google and especially Facebook. Should I not feel a little icky about using React.js?
People are imperfect, and organizations made of people are inherently imperfect. If you didn't allow yourself to use anything made by anyone who acted immorally, you wouldn't be allowed to use anything.
Should I not use Comcast for my internet if it's the only provider in my area? At a certain point the 'karma' associated with taking a hard-line on a company's ethics is no longer worth the inconvenience.
I don't think there's anything wrong with the team behind it. The reason I might choose not to use a technology owned by Facebook is that my use would be promoting it; the more popular React becomes, the more encouraged people are to contribute code to it, which helps Facebook in a minuscule way. Nonetheless, the popularity of React does contribute to Facebook's success in some way.
Comcast is a different story, as I'm not aware of people helping them for free by using and developing their own tools. I get what you are saying by generally objecting to Comcast and facing the dilemma of paying them when you have no other options. I guess it depends on how much you dislike Comcast and if you are willing to replace them with free public WiFi and Netflix. At least with Facebook, there are plenty of alternatives(as well as ripoffs of React).
Contributing to firefox, openssh, wikipedia and the linux kernel also helps Facebook in various ways. Should we stop doing that?
React is an open source project. Facebook did right by it (unlike other open source projects coming out of FB such as Flow). I find this hardline stance kinda ridiculous when it's one of the rare good things to come out of Facebook. It should be encouraged if anything. But even if not, it is a standalone project and a very good model example of how companies should release stuff as open source, so not contributing to it "because it helps Facebook" is completely missing the point of what open source is.
It's a lame argument that just because an evil company like Google or Facebook or Microsoft made an open source product that you are using, you need to shut up and stop criticising them!
The fact that someone uses their open source product is an endorsement to the open source movement, not of the company.
The fact that a public for-profit corporate made a product open source doesn't mean that they have suddenly become altruistic - it just means that it made business sense for them to make it so.
The fact that it is open source means that the company doesn't own the product nor can it control you. And the fact that it is open source means that the product is being made by contributions from the general public too, and not just the company.
I don't think that's what op wrote.
It doesn't make sense that folks use open source code from these companies but then there's shock and outrage that these companies are also using open source code written by others.
>because an evil company like Google or Facebook or Microsoft made an open source product
It's hard to take you seriously when you blanketly dismiss companies as being "evil" (whatever that means). By what metric are these companies "evil"? Wow.
The fact that someone uses their open source product is an endorsement to the open source movement, not of the company ... The fact that it is open source means that the company doesn't own the product nor can it control you. And the fact that it is open source means that the product is being made by contributions from the general public too, and not just the company.
> The fact that someone uses their open source product is an endorsement to the open source movement, not of the company ...
Terribly ungrateful of you, but okay, it's not a crime.
> The fact that it is open source means that the company doesn't own the product nor can it control you.
No, releasing open source code doesn't imply signing away copyright or ownership. For instance, Mozilla owns Firefox (not only in the trademark sense); you can build another browser with their code, but you need to call it Iceweasel or something else.
> And the fact that it is open source means that the product is being made by contributions from the general public too, and not just the company.
Open source doesn't mean that at all. Good old immutable tarball dump can be every bit as open source as accepting contributions from everyone on the planet. In fact, check out esr's The Cathedral and the Bazaar.
"Open source" is already intentionally or unintentionally skewed to represent many things it isn't, please don't add more arbitrary attributes to it.
Companies are not people. It is nonsensical to say that someone is being grateful or ungrateful to some abstract legal construct.
If the executives of some company decide to let their employees work on open source projects, it's because they think this will improve profitability in some way: advertisement, attracting talent, benefiting from free work, facilitating adoption, etc.
Most of these companies benefit from open source way more than they contribute. Which is fine, but let's not assign human emotions to things that are not human. Being grateful to a corporation is the same as being grateful to your coffee machine for making coffee when you press the button.
I am very grateful to people who contribute to open source on their free time.
How could I not? It's one of the most repeated clichés in the history of clichés.
In any case, I am not denying mutual benefit or trying to forbid anyone from anything. I am just saying that if you are shouting "you ungrateful bastard" and the target of gratitude is a legal entity, you are just being silly.
It's the other way around, if you hate them and criticize them so much, you need to stop using their products.
Yes. And it is the western corporates that have deliberately created this anti-competitive business model. This "free to use for your personal data" model ensures that any potential competitor has a very high bar to entry, and it ensures that their product will always be inferior because of the lack of data that can make it better.
You sure it's not the same cheap consumers that cause race-to-the-bottom business models in a lot of other industries too (airlines)?
Right now the only way to charge the average consumer is through their debit or credit card. This makes micro transactions through the web infeasible.
Imagine a world where there weren't such massive legislative moats and selective enforcement in favor of big financial players and against the small.
Personally I believe if setting up the competetion to MasterCard/Visa weren't made virtually impossible via legislation, we'd see many card with transactions fees of $0.000X instead of $0.xx/transaction + Y%/transaction + minimums.
I think consumers would be much more ok with paying for reading a news article at $0.01/read. But alas, such options aren't on the table. I do concede it's hard to prove the reason for current transaction fees is what I attributed in my post, but there is evidence for it.
I also think that the way money is created an placed into the system could be done in a way that created more smaller competitors instead of such large pools of capital, which can more easily try and engage in anti competitive behavior.
I think you'll find that very few people are willing to engage in any degree of per article mental gymnastics to read a news article, and will instead be very very selective about what they read, with ruinous effects for local news.
If Spotify was not $9.99 flat fee you would likely not happily leave it on all night or use it as background music for fear of what it was costing. The difference between the way people used metered and unmetered internet was huge. Charging any kind of fee on a per article basis would have profound impacts on the way people consume media.
If only a few of the paywall newspapers would implement a direct per article fee, even at $0.10-0.50 I would pay for many articles because I'm too lazy and don't want to pull up a javascript disabled google cache version of the page or fire up a VM... or any of the other techniques SWIM would use to get around paywalls.
Facebook tracks you regardless of whether you have a Facebook account.
Developers made these products. I'm sure they would be just as happy getting paid the same wage and making awesome products without working for a morally bankrupt company, but those opportunities don't exist. FAANG are the only institutions on the planet that have the resources to fund large-scale open source projects.
It's not their code. Linux didn't come from a company for example. The list goes on.
The amount of code they give back is peanuts compared to the amount of code they benefitted from to build their businesses. No need to feel guilty whatsoever to use it. Also totally fine to point out the parasitic players.
>Also totally fine to point out the parasitic players.
Who are the parasites?
These bad-faith posts that are really just unconstructive rants are turning this site into Reddit.
He currently does.
There can exist a mutual beneficial symbiosis between companies and the open source community. But at some point the symbiotic behavior can become parasitic, causing harm. I think it's super important when such behaviors get pointed out (here at HN).
The people at Google who contribute too, and champion open source are probably not the ones who are making the decisions that we see as "evil."
Google is competing for top talent engineering, and Googles commitment to open source boils down to a company perk meant to attract and retain top engineers.
Google is a advertising company. Everything else they do is in service of that.
From a consumer perspective, this is not particularly true, but it's definitely a factor in their behaviour. For instance, if Google made all its money by indexing data for big companies and selling the index, do you think they would own YouTube or have created Gmail?
The Ads org within Google is large but fairly compartmentalized and runs like an absolute machine -- there is no questioning that.
What most outsiders don't see, though, is how much independence all the other orgs have. For example, the fastest growing collection of stuff are the Hardware & Cloud areas, which account for probably 35000-40000 employees and are almost completely autonomous. If Ads show up on Google hardware products (let's use Google Home as an example), that's because the Home team exposed a service that allowed the Ads team to do it, not because the Home team was necessarily thinking about display/search ads during product design. In fact, there are many services Google provides where it's well understood that users will not tolerate ads at all (like Google Assistant surfaces).
The reason I responded to the parent was because they made a blanket statement, which, while true if you only look at the fiscal reports, is untrue in practice. One could certainly argue strongly that Google's inability to strike it rich with anything besides Ads is 1) the reason the stock has been in the [relative] doldrums for the past year+, and 2) why Google continues to invest in all sorts of things (like Wing, or Chronicle, or Stadia, or ...) that are pretty far afield, not to mention the pure & applied research happening inside the RMI/Brain org that's often much more academic than practical.
Besides search & mobile (admob) advertising, the only two major product areas inside Google that take a strong stance supporting the advertising mission are Youtube & Maps, but even then Maps tries to ensure it's surfacing useful, actionable, timely, and local information, which generally place it in a different category. Maps also have strong potential via partnerships for things like ride services, appointment bookings, travel, food delivery, and the like... some of which might technically be considered advertising but which manifests more like functional partnerships.
Regardless of my diatribe, the business decisions are what the HN community rails against, and those are typically not the decision made by the rank-and-file (who are likely contribute the most to open source).
"We had horrible gravel roads and the dictator came and built good roads we use even today". Is my late-grandparents argument as to why dictatorship is by far the best way to rule a country. And I 'd happily drive on those roads to tell them they are dead wrong.
Pablo Escobar donated tons of money to the Medellín community. Doesn't make any criticisms against him hold less water.
okay.jpg
Some people don't use FB.
Some people would say FU to the thought of paying $2-3 to FB for anything.
There's no logical contradiction in attacking a company for doing bad things with the open source software they wrote and instead praise company who use that software for good. Software is not good or evil, it's how the company/people that use it that determines good vs evil.
The alternatives have all been found wanting: "intellectual property" (copyright and patents), patronage, religious commissions, advertising.
Busking, performance, physical media sales, merch, loss-leaders, widget-frosting, service/support, and shingle-posting all bypass the HFZMC problem by monetising some alternative good, product, or service.
Several of those specifically tie network-effect, rent-generating monopolies with FCZMC goods. Taxes socialise this process.
What to tax is the next question. Adam Smith preferred wealth and land (means of production).
https://en.wikisource.org/wiki/The_Wealth_of_Nations/Book_V/...
You could GPL it...
Step 1: Launch company to take advantage of the countless hours of free labour provided by the open source community around your project, hoping to get rich in an IPO.
Step 2: Be shocked when Amazon also takes advantage of the countless hours of free labour provided by the open source community around your project to get rich, only they also throw a bunch of expensive developer time at it to make it really shine.
Sometimes the "open" route works well for founders - see RedHat, but that's not easy. Prior to AWS there were also others who were trying to earn money on someone else free product. In the pre-cloud era those "unfair" competitors were easier to beat, AWS is indeed a game changer here.
Maybe there needs to be a new software license. One that allows for commercial use when the project is included as a module or infrastructure but excludes it from being billed separately. So you can use an open source component as a language library, shared object / DLL or even stand alone daemon / service installed as part of your web application (for example) but you cannot make that open source project a product in its own right. Obviously a lawyer would need to write the terms rather than an engineer like myself but I wonder if that kind of software license would offer a happy middle ground?
Way back when I was starting as a systems engineer at one of them-thar unix workstation companies, my mentor in sales laid down a bit of learning on me.
He said "anything you give away for free has no value." His context was that here we are selling big expensive hardware, and while the software is cheap, it wasn't free. I wanted it to be free so as to lower barriers to use the systems. I wanted dev environments to be free, and pay for support if you need it. Again, to increase usage.
You can see similar concepts in what companies decide has no value, by what they are willing to bundle for "free". Cell phone plans in the use are doing 2-for-1 deals of some sort. The carriers believe their services have value, and the phones are merely a means to deliver their value.
OTOH, the cell phone vendors believe their devices are the value, and try to give away software they think will make their devices look better.
Yeah, I know, you pay for it in the end. Costs are always borne by the customers.
Amazon is leveraging its market dominance in ways that are hard to view other than as anti-competitive. I do suspect that eventually, they will be held to task for that. But that they will likely keep operating in that manner for as long as possible.
Azure, with its shared success model, may actually manage to convince OSS projects to work more closely with them. It would not surprise me in the least to see licenses rewritten to be friendly to the shared success model, and hostile to the do whatever it is you want model.
Which worked for a while. You either didn't need the extra tools (portable code was a subset of K&R anyway), or you had ample money to pay for extra tools. (Research&education appreciated that the majority of GNU and other open source software favored Sun's bundled compiler, over other workstation platforms. Though there was some brief upset when a competing workstation line would leapfrog performance briefly (e.g., HP 9000/7xx, IBM RS/6000). And the commercial workstation independent software vendors sold such high-ticket software (e.g., 5 figures per seat, plus support contracts and training), and were already dropping huge money on engineers and hardware, so paying for software tools was a relatively small additional expense. The ISV might also go pay others for seats of third-party tools for developers, like Saber-C, Purify, FrameMaker, Interleaf, and niche developer software like EDA or CASE.)
Meanwhile, GCC was starting to get as credible on some hosts/targets as the workstation vendor compilers (vendor C++ compilers sometimes had bugs). Before that finished playing out, I moved from workstations (I worked on almost all of them, plus the Cray S-MP) and sometimes Windows NT, to move fully to GNU/Linux on mostly PC hardware.
(Edit: I got access to these toys as a super-nerdy kid. I'm well aware that HR screeners shred the resume of anyone who was paying attention when the Web boom started. :)
Oracle closing up “Open” Solaris was a real eye opener for me and I find the licensing / business politics of open source to be fairly interesting.
This is not to say that Elasticsearch is great or had the right response, but it is to say that the effects of Amazon will be and are harmful to the open source movement. I do not see easy answers to how fix this, but I think recognizing this as a threat is important.
1. Amazon takes from open source just as much if not more than any of the other tech giants, in fact the vast majority of managed services on AWS are open source projects with DevOps from amazon.. which by itself is fine but when you consider how much they give back? Fuck all. They haven’t contributed any major framework, system, project, or library that I use. I use things from Google every single day, from Kubernetes to Flutter, to Tensorflow the list goes on and on. I use things from Microsoft every single day in my development from Visual Studio Code, to .net core. I use things like react from Facebook every day. Maybe I’m living in a bubble but Man, this feels like theft these guys take everything to make Bezos richer and give us shit.
2. Although not directly related to Open Source, it reinforces the predatory tactics, Take a look at what they do to Brands on their market place. Any well selling product becomes an Amazon basics, generic, of which they make lots of money and shut you out using Sales data and other predatory practices.
3. You have a streaming service on AWS, an e-commerce shop, any other number of SaaS services? No problem as soon as it’s proven Amazon will launch its own product to compete with you. Everyone told Netflix it was a terrible idea to enrich Amazon but they did it anyway now they have original shows on Amazon video to contend with and I bet you they wish they didn’t.
The list just goes on and on and on. Not saying the other tech giants are angels but man these guys are egregious in the way they take. I don’t blame these companies for crying foul none of this behavior is Moral, it might be legal for now but even that time is coming.
Anyway I know I’m coming off as harsh and you don’t have to agree with me but this is just my two cents. I believe Amazon is the most predatory, largest, capitalist company in America today.
And I'd echo the sentiment of others here - Elastic itself has benefited GREATLY by building their product on top of many other open source projects. Java, Lucene, Netty, etc.
Here's a partial list of dependencies:
com.carrotsearch.randomizedtesting
com.fasterxml.jackson.core
com.fasterxml.jackson.dataformat
com.fasterxml.jackson.jaxrs
com.fasterxml.jackson.module
com.github.spullara.mustache.java
commons-codec
commons-io
commons-lang
commons-logging
com.nimbusds
com.sun.jersey
com.sun.mail
io.dropwizard.metrics
io.netty
joda-time
junit
net.jcip
net.minidev
net.sf.supercsv
net.shibboleth.utilities
net.sourceforge.csvjdbc
org.antlr
org.apache.hadoop
org.apache.httpcomponents
org.apache.james
org.apache.logging.log4j
org.apache.lucene
org.apache.pdfbox
org.apache.poi
org.apache.santuario
org.apache.tika
org.bouncycastle
org.carrot2
org.cryptacular
org.hamcrest
org.jline
org.locationtech.jts
org.locationtech.spatial4j
org.openjdk.jmh
org.opensaml
org.ow2.asm
org.slf4j
org.yaml
ua.net.nlp
All our cloud providers are currently in growth mode.
What will happen when they're in 'mature industry with trapped customers mode'?
We know how Oracle handles that situation.
Not only do we have a long time to figure things out but it's never been easier to be on prem if you need to be.
If there's any takeaways that I have from this journey it's that Azure's platform is simply not competitive and is likely to continue to not be.
Why should Amazon not be allowed to sell its own products on its own site, but the grocery store is allowed to sell its own branded milk at its own stores, in competition with other milk sellers?
The fact that Amazon is rolling open source products into its own offerings isn't anti-competitive: it's a reflection of a key flaw in the open source philosophy. The moat of open source is that it is expensive to maintain and difficult to monetize, thereby disincentivizing unserious forks: but now that Amazon is monetizing it, it's a problem, because their resources are effectively unlimited and they are effectively monetizing it. The potential was always there for anyone to do this, but the difficulty of it meant that no one ever tried.
It creates a conflict of interest when others want to host it, since it takes their revenue away. However, in my opinion anyone should be able to compete on hosting/support.
This revenue goes back into funding development, but places restrictions on how the software can be used - making it "Open Sourced, Closed Service Based Hosting" or something like that. Inotherwords, we need a new name for this type of software.
My question is - if AWS simply made it easy to spin up an elastic search cluster, but didn't offer a specific API/Service - whats the difference here? Very little.
One of the main things Amazon has going for it is that one can usually expect to see the whole gamut of competitive products, read helpful reviews, etc.
But when products are missing from the results for competitive reasons, the trustworthiness of the platform goes out the window.
So this may be a short-term beneficial strategy for Amazon but will surely backfire in the future. I'll be reluctant to buy more stock until the impact of the strategy becomes more clear.
Remember MS and IE? Something needs to be done asap.
Ground Station
Robo Maker
DeepRacer
Bunch of storage features (many could be standalone startups)
Bunch of IoT stuff (same -- SiteWise, ThingsGraph, etc.)
Many launched with real customers, too.
It is NOT EASY to bring a software MVP to market and AWS does it massively, across many efforts, in parallel, EVERY SINGLE YEAR.
Being a developer right now and complaining about AWS cannibalizing open source is a lot like running a late 1400s monastery and complaining about how Gutenberg Bibles are displacing hand-written Bibles.
Naive question: Why not change the license so cloud providers have to cut a % of those fees with Redis?
They did it with ElasticSearch
https://aws.amazon.com/blogs/aws/new-open-distro-for-elastic...
And Java
https://aws.amazon.com/blogs/opensource/amazon-corretto-no-c...
And something similar with Mongo DB.
https://aws.amazon.com/blogs/aws/new-amazon-documentdb-with-...
They can negotiate a special license just for them.
This is basically the middlemen analogy in commerce which internet completely decimated by allowing manufacturers/retail to go directly to users.
I don't feel sorry for them.
This is why I like the GPLv3 license. Sure it excludes my work from ever being considered in a large corporate project but I'll never have to worry about being beaten over the head with my own product by one of them either.
What about all of the contributors to Elastic that are not employees. Why does Elastic own the monetization of other people's work?
Many of our engineers, such as Robert Muir and Mike McCandless, were already contributors to ASF projects before they joined AWS. Robert worked with the Apache Lucene and Apache Solr projects as part of his role at AWS, which led to three more employees earning committer roles on these projects. Mike continues to be a very active contributor on behalf of Amazon to Apache Lucene, the storage engine that powers Solr and Elasticsearch.
OTOH, AWS pretty much xeroxed the elastic stack and added a paywall (at least after a year). Compared to that you still get immense value out of the elastic stack for free.
When they decided to implement features that were only available in the proprietary version of ElasticSearch, they open sourced it.
Charging for infrastructure and management of said infrastructure is not the same as placing a paywall on software.
You can't pretend that a monopoly plays by the same rules as everyone else. For capitalism to work no one can be allowed to control a market influencing chunk of it.
Not true. Even GPL code is able to have proprietary modifications so long as the code stays in-house. As a cloud provider, Amazon can do an end run around the license by simply providing access to your code (with their proprietary modifications) via Web APIs.
That was the root of the whole MongoDB debacle, and several minor skermishes with Redis Labs.
[EDIT:] I thought the "MongoDB debacle" was the wide-open default auth thing? I don't see how that connects...
AWS has never used the AGPL licensed MongoDB code as part of a service, therefore the license change from AGPL to SSPL has nothing to do with AWS. There is no "Mongo fork" as you stated above.
See more in MongoDB's FAQ about the triggers for changing the license, in particular "international cloud vendors". https://www.mongodb.com/licensing/server-side-public-license...
Surely, more competition is actually good for consumers, not bad?
All of this is irrelevant to your argument.
> and then sold access to it.
Are the purchasers of this free software foolhardy, or is Amazon providing them value in some way? I suspect it's the latter.
That's ridiculous. Of course they were keenly aware that they could be competed with with the free software they put out.
They just hoped that no significant competitor would risk upsetting the balance of the open source community, or that they could at least make good money until someone did.
In the meantime, they selfishly avoided licensing with GPLv3 (which would have reinforced the things you care about) because it would have decreased their revenue (which is the thing they care about).
This is almost completely wrong.
Amazon sold a managed service for the project long before they forked it. Elastic.co didn't like that Amazon was doing so successfully, so they tried to restrict access to the open source project.
Amazon, on the other had, offers complete access to their fork -- https://github.com/opendistro-for-elasticsearch