hckrnws
Open Source Software has a hugely important factor that we don't talk about enough, IMHO: it promotes standards that are actually accepted.
Think about it. A new image compression standard comes out. What are the chances that it becomes widely accepted and used if there isn't any Open Source (or at least free) tooling to compress and decompress it? And just "free" might not be enough, because lots of folks need to incorporate it into their own software (image viewers, browsers, chat clients etc.), and they wouldn't take that risk if they could link it, but maybe not apply a security fix if the vendor didn't react quickly enough.
Image formats are a relatively benign example, but deeper in the stack, non-interoperability becomes much more of an issue. How many programming languages without Open Source implementations became really successful? How about OS APIs? Do you think we would have somewhat compatible libc and syscall landscapes without Open Source BSD distributions and Linux?
And then there's version control, implementation of communication standards (bluetooth, USB, but also TCP/IP stacks).... the more I think about it, the more horrifyingly fractured the software world would look without OSS as a unifying force.
> Image formats are a relatively benign example
GIF was, despite patents a dominating image format for a long time and PNG took quite a while to replace it.
If we look at video codecs "open" variants are quite a small part. Heavily patented MPEG dominates the space over webm etc.
> How about OS APIs?
Windows still got a very successful proprietary API. (Or set of APIs)
Yes, open standards and implementations are relevant, but it's not that clear win.
> GIF was, despite patents a dominating image format for a long time and PNG took quite a while to replace it.
Yes, and it caused quite some legal grief for many software vendors.
> If we look at video codecs "open" variants are quite a small part. Heavily patented MPEG dominates the space over webm etc.
We were talking about Open Source Software, not patents. There are lots of OSS codecs for the various MPEG formats.
> Windows still got a very successful proprietary API. (Or set of APIs)
Depends on what you call "successful". Sure, it's been in use, thanks to a quasi-monopoly. But I'm not aware of any other OS implementing the Windows APIs because they think it's a great set of APIs. There are some projects like "wine" that try to implement it to offer compatibility, and it took them many years to make it mostly work.
Would it have been better for society as a whole if Windows had implemented System V APIs instead of their own, proprietary APIs?
Monopolies are the other avenue to introduce new protocols and formats, but apart form that, OSS seems to be the main pathway to introduce them.
> Depends on what you call "successful".
Could you give a reasonably mainstream definition of “successful” that would exclude Windows?
https://reactos.org/ implement some of the windows API
GIF was fully open and free, before it wasn't. After it had established itself as a standard, was when Unisys and IBM started in with the patent they had on the compression algorithm. Note that GIF could have switched to a different compression format, but that would have had the same issue that PNG had in that new images wouldn't be compatible with existing viewers (and that was in the days before the concept of "evergreen" web browsers -- they didn't get automatic frequent updates).
And the other thing holding back the switch to PNG was that people weren't all that confident that there weren't any unknown patent landmines in that standard. You don't know until either one goes off (some patent troll convinces a court that a generically worded patent applies), or until enough time elapses (20 years) that all potential patents would have expired.
And Windows now has WSL to implement the open API, because it's so popular. And implements SSH and sudo, because they're so popular. Etc :)
It goes the other way too: standards promote open source. With standards, there is less of a moat protecting vendors from free competition.
We've seen this play out with compilers. Who is going to buy an expensive compiler when a reasonably performant free one implementing the same standard is available?
Another factor is transparency.
FOSS software is independently auditable without permission. It does not always happen (few things have "many eyeballs" on them) but it is at least possible. AT the very least, it is a deterrent to malice because of the greater risk of being caught.
Yet another is the reduction in administrative cost because of the simpler licensing requirements.
Another is the the huge difference to distribution between cheap and free to use because there is not payment mechanism needed. it makes trying software easier too.
> More precisely, our results show a cost of $4.15 billion if society had to replace these packages once (e.g., OSS still exists, but all of the most widely used packages were deleted and had to be rewritten)
I realize estimating anything numerically here is tough but that number just seems low by an order of magnitude or more. $4 billion is less than the cost to build one aircraft carrier. It's 1/3 the cost to build the Abraj Al Bait tower in Saudi Arabia. It's like 3% of the net worth of Bezos, Musk, Eliison, or Zuckerberg.
Recreating Linux alone would be a massive endeavor, taking hundreds of people many years, to say nothing of gcc and clang and git, and those are just the developer tools needed to build all the other stuff. Estimating $300,000 per developer (including salary + benefits + office space + computers), $4 billion is 13,000 person-years. Google has about 26,000 developers, so it's saying Google could recreate all open source software in 6 months. Or one tenth of Google could recreate all open source software in 5 years. Not that I have a better way of estimating, but the number doesn't seem realistic.
Firefox alone is like 21 million lines of open source code and I wouldn't be surprised if Chrome was similar. Microsoft, with hundreds of thousands of employees, may of them developers, chose to abandon their long-established browser codebase and hop on Chromium because maintaining a browser engine, probably about 80% of the code, was just too much for them to handle. And that's one single app.
Simply maintaining Firefox takes a couple hundred million dollars a year, and if you scaled back to minimal effort outside of security fixes and critical new features to stay competitive, I don't think you could get it under one hundred million dollars a year. Developers are doing a bit more than coding Firefox of course, there's CI and deployment of new releases every two weeks to hundreds of millions of devices that's all automated and requires maintenance and updates bt the point is that if one project, Firefox, costs hundreds of millions of dollars, how can all of FOSS be as little as 4 billion?
> Estimating $300,000 per developer
Or broaden then base of developers available to replace it and you can get an estimate a few times lower, so it'd take ~25 years to recreate it with a workforce 10% of G. Is 25 years more realistic?
Not to mention said company needs to pay for all the tooling from the other proprietary companies going out of business all the time.
The prices with be exponential.
The method in the paper definitely seems flawed: if, for example, PostgreSQL would not exist, that doesn't mean every software company would be developing their own RDBMS -- the much more likely effect would be more mega-yachts for Larry Ellison...
I'm not even sure there is a sensible method to quantify the economic impact of Open Source, since it's pretty much an "alternate universe" situation. Without OSS, Linux would most likely not exist, meaning that... everyone would be paying for Unixware per-seat licenses? Which would probably make AWS a nonstarter, etc. etc. On the other hand, there would most likely be a lot more profitable small software companies, but not enough to make up the difference.
So, "Open Source is a good thing, economically" might be a defensible statement, but putting an actual Dollar value on that will always be more fiction than science.
Yeah, it's weird. "How would people behave if everything was different?"
I don't see how there's either $5 VPS or Heroku free tier in a world without free Linux on commodity hardware, and I'd argue that without those, software would have taken a much smaller bite out of the world than it has over the past 15-20 years.
One thing I know is that all the companies I've worked for in my 10-year career would not have been possible to start the way they did without open source software. First of all, they all had a period of at least a year where they were just an idea and built up an MVP for no cash. If license costs had been significant, who knows how that would played out.
Second of all, throughout my career a significant portion of my coworkers have been self-taught. Actually, everyone has, to some degree, even the people with actual CS degrees have run stuff for fun. And for everyone, that's depended on license-free software, no money to install and free to mess with. If this had come with a price tag or other limitations, maybe we'd've done something else on our Tuesday nights instead.
When I started my career Linux and FreeBSD weren't as good as they are now. People we're not playing around with them nearly as much. So UNIX experience was hard to come by, and the big players like SCO, Sun, SGI, etc were really the only game in town. When Linux got good suddenly finding people with UNIX experience became A LOT easier. This also had the effect of decreasing the value of UNIX experience and all those people who had learned on big company UNIXes were now competing with kids who had learned UNIX systems from messing about with Linux.
> it's pretty much an "alternate universe" situation
Yup. In the early days, hardly anyone paid for software (as such); when you paid for your mainframe, the software, complete with source, was bundled. You also got training, and engineers to help you use the thing. When hardware became cheap, everyone and his dog started writing software, just to learn (or play) with the new hardware.
It's very hard to envisage a scenario where cheap, powerful machinery can't spawn a community of hackers producing free (or very cheap) applications. If your counterfactual scenario contains internal contradictions, it isn't really possible to draw useful conclusions about how that scenario might develop.
Sure, but the point of the HBS paper isn't to paint a realistic picture. It's to imagine a hypothetical world where OSS doesn't exist, in order to derive a monetary value of its worth.
I agree with this article that the actual values might not be realistic, but it's a good thought exercise that highlights the importance of something most of us take for granted. Arguing about the methodology used to arrive at those numbers is missing the point, IMO.
Looking at hardware might be a good way to see how the "alternate universe" plays out. Hardware isn't free and can't be infinitely copied. So if you want to be a self-taught EE there's a certain minimum financial barrier to entry (like you have to buy FPGAs and breadboards and soldering irons and so on). There are far fewer hardware hackers around probably as a result.
This alternate universe already exists. Many companies out there writing software in Microsoft C#, deploying on Microsoft Windows, querying Microsoft SQL Server.
Most common server target for new deployments in .NET is Linux and PostgreSQL is a very popular DB choice to accompany those.
It makes no sense to estimate the total cost of the proprietary equivalent of _all_ that is currently OSS at $177M. It would be spread over at minimum thousands of companies and each company would try to get their margin, needs to be rewarded for the risk they’re taking, etc.
The HBS method to get to 3.5X isn’t sensible (as the author points out, not everyone would build) but the truth is somewhere in-between.
The COGS of software would be significantly higher if there was no OSS. But everyone knows that already. I don’t think any new information has been created here.
I agree, if you think of just a few big OSS project, like Linux, Postgres, Sqlite, nginx or apache web server, run as proprietary companies each of those would easily be worth way more then $177M (for reference, nginx was sold for $670M in 2019).
> It makes no sense to estimate the total cost of the proprietary equivalent of _all_ that is currently OSS at $177M
Right. I mean, I would guesstimate, say, sqlite alone to cost in the ballpark of that amount of money to develop from scratch in a proprietary way.
I mean, working from the initial hypothesis, if sqlite suddenly disappeared and someone starts a company aiming to develop and sell it, how much $$ would VCs throw at the project? How much at Rails? ext4/btrfs/zfs?
The latter gets interesting because Sun threw actual money at ZFS. Comparatively, how much money has, say, Apple spent on APFS? MS on NTFS? How much has been spent on Unix and its descendants?
Either I'm missing something, or that 177M for all of OSS just flat out doesn't add up.
It's in-between, but not in the middle
But if everyone knows, what's your estimate how much higher that would be? M, B, or T?
I think a better estimate of cost would be to take the number of developers, multiple by an average salary, add the costs of marketing folks and management, add the costs of offices and then multiple by 3 (profit margin).
That would give the base a amount that companies building the replacement software for OSS would need to earn.
I doubt very much that would be a figure around 177million.
Also what happens now is that companies take OSS, modify it in house and then either give back those changes or use them as a competitive edge. Would this be possible if the software is sold with a restrictive proprietary license?
>multiple by an average salary
This is such a weird thing to use when on other hand you have developers slinging code for basically pennies and then silicon valley guys making almost seven figures.
There really is no good way to estimate any of this
> There really is no good way to estimate any of this
True that but instinctively, I can't imagine that 177M is in anyway near the ballpark.
> The top 5 programming languages contain C (including C# and C++), [...]
Including C++ in the "C" category is _maybe_ defensible (at least the use cases overlap), but including C# is crazy. And then in the same sentence they list JS and TS as different languages!
Would companies like Google or Facebook even exist if they didn't have access to Linux, GCC, etc. in their early beginnings?
If they could afford the server hardware and datacenters to house them in and the fat, fat paychecks they were giving their employees, they could afford the software licenses.
Everyone here acts as if no capital equipment intensive businesses, like Tesla or SpaceX/Blue Origin, ever manages to get off the ground. Had Linux, GCC, etc. not existed, Sun Microsystems or whoever would have charged whatever the market would bear without strangling it with Microsoft + Intel trying to undercut them on price/performance to keep them honest. For most software businesses trying to get off the ground, software licensing costs would have been a significant budget item but not overwhelmingly so. [EDIT] That's already the case today for per-employee software costs like Windows + MS Office and the zillion SaaSes (Slack, Okta, etc.) that most businesses use.
> If they could afford the server hardware and datacenters to house them in and the fat, fat paychecks they were giving their employees, they could afford the software licenses.
Two things:
1. That's not how it started
2. You underestimate just how expensive Oracle was (is?)
The more hilarious one is Apple.
If open source didn't exist then NeXT would've still been able to build the Mach kernel but would not have the resources for the rest of the stack. So the only option would have been BeOS.
Which means no Steve Jobs at Apple, no iPod, iPhone etc. The whole industry would look very different.
If BSD didn't exist, NeXT would have built on SVR4 like the other Unix vendors. It would have been slightly more expensive.
musing over what might have been seems pointless and the only safe bet is probably that it would have gone a different way
This seems like clickbait, but I do not want to label ideas prematurely.
I understand the point that the author is picking a pretty narrow window of potential benefits from Open Source software.
For me personally, having access to open-source software made it possible for me to learn and apply data analysis and visualization techniques when reasoning about incarceration data. I used the Seaborn Python library (open-source) to help me complete my final project and get my bachelor's degree. With education being a provable form of climbing the social ladder, and open-source software helping me, hopefully, the author can see the value for developers (who will likely go on to work at a company).
...Therefore he decided let the world know of his reasoning
"A single company and 177M$ could rewrite everything that is open source" he wrote, on his 30M cloc open source browser, which dutifully sent packets to his router running 35M cloc open kernel, compiled with a 10M cloc open compiler.
"That figure looks a bit low", said the developer, thinking about all the software in his linux distribution, mostly-open phone and untold gizmos lying around in his house
"Didn't we sell mysql for 1B$ in 2008? Good thing nobody noticed" someone thought in the background
"...You want to do what with that chump change?" thought the red hat executive, a 19k employee and multi-billion dollar company built around open source
but it was all just a passing thought, and everyone forgot about it a few minutes later
Does cloc = Count Lines Of Code ?
I've only seen LOC as an abbreviation, does cloc have a different meaning?
They're probably referring to cloc the counting tool. https://github.com/AlDanial/cloc
And IBM paid $34 Billion for Red Hat
Is the theoretical cost to reproduce what OSS has created even the important metric here? The benefit to society seems like it has been many many times greater than the value of the labor contributed to open source software.
Ah, this is easy, the value of software is essentially 0.
It's possible some one may exhort money via legal machinations but other than that I don't see how it can have any value whatsoever. It's write once then exist infinitely forever, how is that not valued at 0? If all it actually takes is a little time and humans have that in spades.
Software may facilitate doing some other activity more efficiently but without that what would it actually be.
Recent and related:
The Value of Open Source Software - https://news.ycombinator.com/item?id=39340277
> "More precisely, our results show a cost of $4.15 billion if society had to replace these packages once"
Or 1 month of Microsoft Office revenue.
I generally don't accept the result of studies which claim "If we had replaced W with X, we would save Y amount of Z." IMO, there are too many variables involved to make such assertions. The models for coming to such conclusions are way too simplistic.
For the same reason, I don't even buy retrospective arguments like "We switched from W to X and ended up with Y fewer/more of Z."
It reminds me of TypeScript mania when companies where claiming "We switched from JavaScript to TypeScript and reduced bugs by X%."
It's nonsense. They could probably have reduced bugs by X% merely by rewriting the project, even if they stuck with JavaScript. Rewrites are almost always much better than the first iteration, simply because developers learned what the issues and pain points were and were better equipped to avoid them during the rewrite.
This article is just a stealth extension of the so called 'open source sustainability crisis' argument.
Basically, it's just a bunch of people wanting to work on open source software and demanding to be paid for it. Like literally demanding you pay them for using their software!
I'll repeat here what I always say to them. If you demand to be paid for creating open source software, then maybe it's not for you because you don't seem to understand the license you added.
I will always consider OSS extremely harmful to normally employed software engineers and supporting roles around them. OSS makes it easier add value to a company, something only the owners, shareholders and, maybe, founders profit from. It's bad for software engineers themselves, because you need less of them.
Without OSS, there would be much more duplicate work available across companies.
> Without OSS, there would be much more duplicate work available across companies.
That sounds very much like bullshit job + NIH syndrome twisted into an employment rate political play, creating an artificial source of work that duplicates effort and overall wasteful energy expenditure. A close argument to "automation is killing factory jobs".
If you want to create a world where there's no bullshit jobs at all then you'd be creating a world where millions are unemployed.
We are rapidly heading in this direction with automation. AI is beginning to fulfill the final step in this process creating a world where 90% of jobs don't need a human.
> If you want to create a world where there's no bullshit jobs at all then you'd be creating a world where millions are unemployed.
With such absolutes we're entering wild hypothetical territories.
An alternative viewpoint would be that no bullshit job => everyone does a humane job with a purpose. Wouldn't that be nicer than breaking one's body or mind for a soulless job, artificially so just to get money to put food on the table? I think we're fairly happy to not have to hunt for food and plant crops.
A second alternative would be that millions (90% so... billions? all?) being "unemployed" would mean (as you seem to imply) they wouldn't get paid (or they would through some flat social revenue) and money would rapidly cease to have any meaning as we understand it.
Both cases converge to a post-scarcity economy[0], one popular example of which is the Star Trek economy[1].
Of course it's nicer to give everyone a humane job. It's not about what's nicer. It's about reality. There's simply not enough jobs to go around if there are no bullshit jobs. Your statement here doesn't make sense to me. Yeah it's better but what does better have to do with the realistic consequence? Nothing.
Very little jobs for people with sub 80 IQ which makes up a significant portion of the population. If any of these people hold jobs it's most likely bullshit.
If millions are unemployed that doesn't mean they won't get paid. But if they didn't it doesn't mean money has no meaning. I didn't mention the consequences of that. You did.
All I said here is that if we get rid of all bullshit jobs. Millions will be unemployed. What that means is up to you decide, I simply stated the immediate realistic consequence of no bullshit jobs.
Lots of assertions without evidence. We can organize work how ever we want to if we have the political will. Your entire line of argument reads like something from a lunchroom chat in high school.
Attacking another user isn't allowed on HN, regardless of how wrong someone is or you feel they are. We have to ban accounts that post like this, so please don't. Your comment would be fine without the last sentence (and arguably the first one as well).
If you wouldn't mind reviewing https://news.ycombinator.com/newsguidelines.html and taking the intended spirit of the site more to heart, we'd be grateful.
And You're fucking rude. What the hell is up with that Wild u turn there into hostility. You need to fucking tone your shit down.
This is causal conversation not a scientifically rigorous paper making bold claims. That comparison with highschool was Totally uncalled for. Next time either ask nicely for evidence or if you want to be rude please do so somewhere else.
For now we are done. This conversation is over.
Please don't post aggressively or in the flamewar style to HN, no matter how bad another comment is or you feel it is. It only makes everything worse.
Edit: you've been doing this so frequently in other threads (e.g. https://news.ycombinator.com/item?id=39361150) that I've banned the account. If you don't want to be banned, you're welcome to email hn@ycombinator.com and give us reason to believe that you'll follow the rules in the future. They're here: https://news.ycombinator.com/newsguidelines.html.
I learned that OSS is actually worthless. If the code, as an artifact, had any value there would be all kinds of tax implications.
This is a weird point of view to me but still a valid one. I agree with it to some degree: Having the code is not as valuable as the capability (skills, knowledge, etc) to change it. In other words, the most valuable part of an OSS project is not the code but the core developers.
> This is a weird point of view to me but still a valid one.
definitely rather weird than valid
You can't tax friendship and love, are those worthless?
That's not how taxes work. Maybe avoid spouting opinions in areas where your knowledge is so obviously weak.
Crafted by Rajat
Source Code