Bruce Perens: Building a 'billion dollar' startup with Crystal and Lucky [video] | Hacker News Reader

Bruce Perens: Building a 'billion dollar' startup with Crystal and Lucky [video]

youtube.com

45 pointszdw62 comments

Well, the audio video quality is terrible, but the message -- in the second half, you have to power through a bunch of stuff about Crystal and Lucky -- is very interesting.

He describes his vision for "post open source" license, which he is currently developing. His goals seem to be to to empower software developers to take back power from megacorps which have in his view subverted the nature of open source and turned it into a "resource extraction" scheme.

I think there is a struggle between open source, having a viable business, and having a VC backed business. Folks aren't working that out well. VC backed businesses are different from viable businesses in that a viable business needs to be able to make money to pay expenses and operate while VC backed businesses need to make returns on investments for the VCs.

For open-source projects that are the top level (e.g., PostreSQL) and being paid to work on it... it's often useful to have an enterprise market and support model.

Side note, open source works better business model wise when things are more distributed. More centralized with megacorp clouds is harder for open source based businesses.

For libraries that can be used by many projects, open source is a great way to go. I've done semver, vcs, and other libs this way. They aren't the business thing being sold but enablers of other things being build.

Interfaces are great and we need more standard interfaces. They make is possible to have open source and proprietary solutions that work with the same stuff. Businesses can compete and work with more freedom to do so with open interfaces.

Businesses trying to have an open source software SaaS that they run in public clouds are always going to be at a disadvantage. The public cloud provider can always do it cheaper and undercut the business. Businesses that try to go that route will end up having business issues.

Physical products are another space all together. It's the thing (hardware + software) that make it work. Open source software is great there, too. I see it in the car I drive.

Where we license things as open source should be coupled to our thought out business models. It's not an open source or post open source world. It's about the right tool for the job in front of us. We need more thought and talk on that.

Another way of saying "post open source" is "not open source", or "proprietary". But, of course, those wouldn't get as much traction as trading on the good name of Open Source.

We already have some copyleft licenses that prevent the kinds of proprietary SaaS usage that have prompted recent complaints. People and projects don't use those licenses; they use permissive licenses, and then get surprised when companies use their software under those permissive licenses. I've even seen people complain that if they use copyleft licenses, large companies won't touch their software. That's entirely the point! If you want companies to pay for an alternative license or an exception, you have to choose an Open Source license that they're not already willing to work with.

Before we even consider giving up on the Schelling point that is Open Source, perhaps we should make better use of the full spectrum of Open Source licenses we already have.

As someone that has watched the evolution of open source, it's baffling to see these discussions. Developers adopted those licenses in many cases because they wanted their software to be used in the enterprise. Now that the "permissive" licenses aren't working out, they want to scrap open source altogether.

They were told about the devil's pact they were doing regarding adoption, but since they were the same group that believed in company propaganda like do no evil, they went ahead anyway.

That’s a really uncharitable interpretation. Permissive licences were not a problem in the internet economy of 20 or even 10 years ago. The challenge now is that a very small number of internet companies - principally Amazon, Google and Facebook - have such market power that they can effectively co-opt permissively licensed products for their own ends. That wasn’t the case 20 years ago and probably not 10 years ago either.

In the golden age of comercial software you still could get the source, however it was part of the commercial version not the shareware/demoware one, or as additional option.

Naturally with a license that only allowed its use on the context of understanding the product, e.g. source code for the C and C++ libraries of a compiler for use in debugging sessions.

Corporations legal departments just found out a way to use non-copyleft licenses for the 2nd coming of shareware/demoware, while cutting down development costs in the process.

It's evolution. When we started Open Source, the corporate adoption we have today was inconceivable. The problem is much larger than just not liking a permissive license. It might be clearer if you listen to the talk.

He knows we already have some licenses like that. He mentions that the megacorps that have come to dominate open source use their influence to give the worst licenses the highest profile. This could be arguable, but I think it makes some sense and he did address it.

Starting a new movement with a new license could be a way to escape the current dynamics.

> He mentions that the megacorps that have come to dominate open source use their influence to give the worst licenses the highest profile.

That's certainly true, but that's not a good reason to declare bankruptcy and throw away the whole movement.

> Starting a new movement with a new license could be a way to escape the current dynamics.

A new movement seems more likely to end up worse, by not maintaining compatibility with the definition of Open Source; that definition exists for a reason, and its requirements don't just facilitate participation by megacorps, they facilitate participation by everyone. Just about every new license proposal I've seen has failed to actually be an Open Source license. "Hey, as long as we're changing the requirements, let's just go full 'non-commercial use only'", or "Hey, as long as we're changing the requirements, let's try to define ethics in a legal document". All the same mistakes over again that people had to fight to reject the first time around.

We don't need a new movement. We might need an improved license that's still Open Source, and a better marketing plan around that license.

Actually, I am well aware of your concern and thus called it "Post Open". Everything that promotes it is very clear that it is not Open Source.

That phrase, much like the software and licenses it describes, is not an improvement.

> His goals seem to be to to empower software developers to take back power from megacorps which have in his view subverted the nature of open source and turned it into a "resource extraction" scheme.

And he's going to achieve that with a software license?

It is possible, but I doubt it will gain any major adoption.

"subverted the nature of open source and turned it into a "resource extraction" scheme. "

I'm not sure if that's a fair statement because the notion of 'resource extraction scheme' is entirely subjective.

If people are using the software in accordance with the license that you afford it, then that's it, there's nothing more you can ask.

If it turns out that 80% of the value is going to be captured by 'Big Corps', well, then that's what it is and it's entirely up to the Open Source developer to contemplate why they would want to do this or not. To each their own.

I suggest however, that there should be a better 3rd option that frankly doesn't exist, which is to allow devs to get paid commensurate with the popularity of their software, which would technically be commercial, but wherein the terms would actually be fairly open by any reasonable measure.

I suggest this dichotomy between 'Stallman/GPL vs. Evil Corps' is completely false and that it's mostly grey in between. It's just that there's a little bit of a cliff between the more open license and harder commercial terms which make things quite difficult for everyone aka imagine your corporate lawyer asked to review every one of the weird commercial terms of the 300 or so Node.js packages you're using ...

Which is why we are basically back to the 80's with commercial software and demo versions.

With the difference that demo versions are no longer time limited, and even with the demo version there is code available instead of being an option on the commercial product.

Companies just call it open source instead, regardless of what is stated at OSI website as definition.

> If people are using the software in accordance with the license that you afford it, then that's it, there's nothing more you can ask.

Actually, I like people to do more than the bare minimum that's legally required of them. Unfortunately corporations usually don't do that.

Being a 70's kid, the current state of open source feels very similar to the 70's freedom movements, where a large majority ended up working in offices Wall Street style during the 80's.

Maybe there is a lesson there regarding human behaviour, and attitudes regarding the need of income in capitalist societies.

FOSS used to be about developers scratching an itch but now it's about developers who are itching to get scratch, hohoho.

(For non-English speakers, "scratch" was formerly popular slang for money in the US.)

To keep with my analogy, if you aren't on a community farm to live up with the ideals and show it to the man, eventually you will need to scratch to get something from the supermarket.

Sorry. I got a new headset after recording that.

LOL, thanks for taking the time to read my comment and feedback :)

It seems to be at about 7:51 that he starts talking about the problems he sees with current open source. Here's a link to that point in the video [1].

Around 12:47 is where he finishes that and concludes we cannot fix open source, and talks about "Post Open". Here's a link to that point in the video [2].

[1] https://youtu.be/XRl-it1-ruI?t=471

[2] https://youtu.be/XRl-it1-ruI?t=767

This kinda seems like it's creating a tweaked ASCAP or BMI regime for "postopen" developers. I'm not sure what to think of that at this moment.

Your regular reminder that there is already a permissive licence that effectively prevents use by megacorps: WTFPL. Google won’t touch software licensed with it and Intel have also refused it in the past. And yes, you can add a “no warranty” disclaimer in your readme too.

> a permissive licence that effectively prevents use by megacorps

_Why_ do mega corps like Google and Intel refuse to use it? There's nothing on Wikipedia [0], and the license is extremely short and clear.

From your description I expected a license that explicitly denies permissions for certain purposes or types of companies. It's very much the opposite of that! :)

[0] https://en.wikipedia.org/wiki/WTFPL

If it's not meant more as a joke than anything else, it should be.

Can someone share their opinion about Crystal? Is there a legitimate chance it will be used by startups in the future?

The other part is interesting, but Crystal got my attention a lot

Ruby is still very popular, and Crystal's syntax is heavily inspired (i.e. very familiar) by Ruby. This is similar to Julia's approach of making the language syntactically familiar with MATLAB.

If a bunch of startups continue to love Ruby's initial expressiveness and productivity (early code write & read cycles), then unless Ruby improves on similar dimensions then Crystal adoption may pick up.

Of course, Crystal's adoption could be stopped if Ruby (and Rails) improves on the dimensions that Bruce Perens mentioned in his video; namely correctness, reliability and efficiency. Supposedly, Crystal is supposed to be faster, more reliable and just-as-efficient-to-write given the syntactic similarities that many developers love about Ruby.

Title doesnt match video title, intentional?

The submitted title (now changed) was probably a summary of what 'zdw thought was the most interesting take-home from the video.

tl;dr anyone? (Pretty please)

I've listened to only 2/3 of it, but: he promotes Crystal+Lucky (crystal is compiled) as good replacements for Ruby and Rails. The second part of his talk is about promoting a new variant of open license called PostOpen, which will require commercial users of Post Open software to pay 1% for using, 1.5% for using without sharing modifications. There is a 10% fee for worse offense. All percentages are percentages of revenue. This is partly aimed at large companies that host open source software with few modifications as a service and charge for it. Money goes to PostOpen and possibly conventional Open Source developers.

Ah, so a linux system contains more than 100 packages, does that mean if it used this scheme companies would have to pay out >100% of revenue?

Colour me unconvinced of the viability of this scheme.

leftpad intensifies

No, and this is expressly discussed in TFV. It's a flat 1% regardless of how many "post open source" resources are used. This is intended to encourage adoption of more POS (unfortunate acronym) once you use any.

Well if people want their ideas to be heard they need to write them down instead burying them in long videos amidst topics i dont care about. (Not blaming bruce, we're not the target audience here, but im still not watching a long form video).

I still think its a weird business model. I can't imagine any propriatary software would work with that. Its very discouraging for small projects that do simple things, or for companies that want to try out software a little at first before committing to using it at a large scale.

I think any license that requires a percentage of revenue is DOA.

First because involving all the chicanery of accounting to figure out my fee is asking for lots of resources just to calculate and audit fees.

Second, unpredictable costs are bad. If my company’s revenue doubles in a year, that doesn’t mean that my department’s budget doubles. Or that I even have enough earnings to cover licenses.

Finally, this is hard enough with a single product. My org uses thousands of products. If they all charge 1%, where does that leave me.

PS- morally this just seems dumb. If my grocery store charged me more or less depending on my income or the value I derive from a tomato, I won’t shop there. Just publish a price and let people decide to buy or not.

That's not correct. It's a flat 1% of revenue regardless of the number of Post-Open software packages used. (From the slide at 13:30 "The same fee whether you use 1 Post Open program or 1000.") The organization that receives the payment is supposed to handle splitting up the revenue among software package authors, so it does require auditing of all usage.

That helps with my final point but #1 and #2 are deal killers as well.

This also places an undue burden on the payment receiver as they have to get into the business of running enterprise audits to find out who is using what.

Yeah, I don't know what corporation would be happy with a license where the cost is nebulous like this.

I would suspect lots of Hollywood Accounting is likely; putting all the PostOpen software in a subsidiary that has no revenue, or developing your own software under PostOpen but not distributing it outside, so that the majority of the usage is apportioned to affiliated companies.

Plus, apportioning by usage is a negative incentive for optimization. If your DB reduces query runtime by 10% in the next version, it reduces its revenue, assuming other PostOpen software is in use and doesn't optimize.

Maybe. That depends if OSS devs switch to this or not. This only hurts those making money off OSS which is fair. Think of it as contributing to improving the OSS.

I think the more software the better so if people like using this then good for them. More software in the world.

But it doesn’t replace OSS and I think will produce different software. I can’t imagine many developers switching to this. I wouldn’t contribute to a project with this license because I don’t want to bother with some incremental level of income. I’d rather just donate time.

You make a compelling argument except for ignoring the fact that many licences (in particular patent licences) work on percentages of sales or revenue. So it seems not to be a problem in those cases, why is it for software?

I’m sure there’s some software that works on percent of sales or something. But it’s not common. I’ve worked on some fixed cost, percent value contracts and they had very distinct parameters unique to the contract and they weren’t very good for generalization (eg, “your foo process costs $50M today, I’m going to deliver changes that reduce that cost, I want 50% of the cost reduction” with tons of pages of caveats and details and stuff).

I’m not saying that PostOpen is impossible or can never be used anywhere. Just that it sucks and is not feasible for broad use.

What if only companies above some revenue threshold were charged? Would you feel differently if licenses had a progressive fee of some kind?

I don’t sell software so this is just theoretical. But I wouldn’t want to know my customers’ revenue. That’s too much work and not my business.

I also wouldn’t buy software that had a cost contingent because I wouldn’t want that kind of relationship with my software vendor. Of course, this happens now with enterprise software where a big company will get a quote for $5 and a little company will get a quote for $1. But having something explicit is illogical since software is a near zero marginal cost product.

But even for real world stuff, I’d never hire a gardener who charged differently based on customers income.

Having spent 20 years in the film industry (see https://www.imdb.com/name/nm0673302/), I am well aware of your concern. The demand is for 1% of end-user revenue collected, through all entities. This is to use all Post Open software, not just one program. Obviously this is not going to gain immediate acceptance by many companies. But if the collection gets large enough, there may be a tipping point.

I see a low probability that anything licensed "post open" will see adoption if there is a royalty owed for it.

Imagine the business decision of selecting between a post-open licensed ware and BSD licensed similar functionality. In return for 1% of revenue and the hassle of an annual usage audit, you get what value in return, exactly? That is the key question.

So as a purveyor of post-open software, you must have a business proposition that closes the deal. Not impossible, but different from the way most OSS projects operate today. Your skepticism is reasonable. To separate a customer from their money, you need to provide obvious value.

It strikes me that once you take one post-open package into your stack, the incremental cost of the next N is zero. So maybe there is enough virality in that feature to drive adoption. One high-value post-open project could create a coat-tail effect.

I just skimmed the video so my TL;DR might not capture everything he said.

Basically, he advocates a new license that he wants to develop that basically improves remuneration for open source developers. Use case: A "Post Open" software is used by a company, then this company has to pay a certain percentage (1 to 10%) of their revenue to the "Post Open" software. If it is using multiple "Post Open" packages this percentage is divided among them (according to usage).

Software in this scheme will be allowed to be modified, redistributed etc. and it will also contain a public API that defines the boundaries of the program (so it's not about linking anymore).

I hope that captures the key points.

Pretty good summary.

Clarification: for-profit users pay a percentage of end user revenue. non-profits and individuals pay nothing. Automated auditing envisioned.

This is a proposal, not nearly ready for roll-out.

The concepts are interesting, but only a rough framework exists at this time. The ideas are worth discussing. I wish there was a succinct blog post or something about them rather than this painful video.

There's a couple of other points of note:

- Usage auditing for and reporting for Post-Open software is required alongside the payment. This is a huge reversal. Not having to do onerous license compliance monitoring was a big selling point touted by FOSS advocates for the past two decades, pointing to horror stories about audits demanded by the Business Software Alliance and huge payments for license violations.

- It's called the "Post-Open" license for a reason: Bruce says FOSS has been defeated. He points out that corporations have both found ways to make money off of FOSS without contributing much and even hijacked the governance of FOSS for their own benefit. This is extremely startling to hear from someone so well known in open source.

What's unclear is how is the revenue percentage split across all of the various contributors of a particular "post open" package. Suppose there are 10 "Post Open" packages; do all of the packages deserve to get an equal share of the 1% rev share? Maybe some "post open" packages are more technically complex than others? Maybe some "post open" packages are more key to the value-add than others.

Even if there is only a single "post open" package which gets the full 1% of the rev share, what about one developer which contributes a whitespace or spelling fix, versus another developer which contributes a key part of the package? What if one developer contributes a huge number of lines of code, but it's for a feature which isn't actually used at all by a particular billion dollar use case of said "post open" package?

The devil is really in the details.....

Already posted 2 days ago: https://news.ycombinator.com/item?id=27961414

It's not a dupe if the earlier submission had no front page time and discussion.

For me it's very simple - when I see GPL 3, I move on - even for personal uses.

I can't imagine why I'd want anything to do with this license either. It's an unnecessary burden. Just stay away from it.

Why is GPL 3 objectionable for personal uses?

The completely serious answer is that some have turned this into a religion. GPL software is simply impure to them, so they won't touch it.

Which brings us to the question: would we be here if there had been no GPL at all?

No we wouldn't.

The UNIX world and Windows would keep taking BSD code, assuming the court case would have ended the same way as it did, and that was it.

The GPL haters tend to forget that GCC was pretty much ignored until Sun started the trend to split UNIX into user and developer editions, and without GNU there wouldn't be a userland for Linus to plug into Linux.

Whatever, all UNIX clones that came after Linux (specially for IoT) are BSD/MIT based, so the long term roadmap is already quite obvious.

My strategy is to ignore the license and do whatever the hell I want. So far I’ve been 100% successful with it.

Reply on news.ycombinator.com