Comment by lazaroclapp | Hacker News Reader

Comment by lazaroclapp | Hacker News Reader

Parent Full thread

There are basically two groups of large software companies around right now: those which make their business by collecting data, and those which make their business by licensing software[1]. The first group has an overwhelming incentive to not support privacy too strongly. The second group has an overwhelming incentive to not allow too much openness. Until a better business model (or zero-knowledge machine learning) is found, no large for profit company can support both goals to their final conclusion[2]. So we are left choosing one evil or the other[3].

[1] Sure, Apple only really sells hardware directly, but the software is a significant part of the reason a lot of people by Apple hardware (e.g. 'Mac's don't get viruses', 'iPhones have a better user experience').

[2] Sure, Google has some significant internal efforts for supporting better user privacy (e.g. https://googleonlinesecurity.blogspot.com/2014/12/an-update-... ) and Apple maintains some superb open-source software (e.g. http://llvm.org/ ). But in the end, Google can't be a "privacy company" without hurting their business model and Apple can't be an "open source company" for the same reason.

[3] Or the non-trivial inconvenience of being a self-hosting free software purist

Actually, there is a third business model which doesn't exploit its users' privacy, nor does it have to resort to closed-source licensing (but selling add-on services like support and customization). This is the model adopted by companies like Red Hat, Canonical and to some extent Google (in a few products).

The future, in fact, belongs to this third business model that helps a business earn profits without hurting its users in any manner. Ultimately, all the IT companies will have to embrace this model in order to stay in the market and survive. Competition will ensure that they will.

The reason I specified large software companies, is that support and customization works well up to a certain scale, but won't give you a "big five" size company. A software Mittelstand which profits from support and customization might indeed be a possible future, but I think is far from a certainty. Also, I wouldn't consider that to be without disadvantages.

Two example disadvantages:

1) As it is, support and customization for specific non-technical paying users are among the things many top engineers least like to do. The reason being that it takes away from time solving the problems of the large mass of non-paying users. Even under the support & customization model versus the proprietary model, the number of paying users is much smaller in the first case in general, which creates a smaller "high priority" class of users.

2) Certain features and applications, such as traffic-aware maps or voice recognition engines are easy to build by huge centralized organizations which hold all the necessary data. They are challenging things to implement for loose collectives of smaller software companies, specially in a privacy-aware way.

I can't upvote enough that excellent summary of the situation of software companies.

One way to solve that would be to have governments support and subsidies open source software development, but I don't see that happening in the next 5 years at the very least.

So, it is far from a simple problem. For common infrastructure, one can argue governments could fund open source in the same way they fund highways and bridges and physical fiber optic links. But I am not so sure that model would be the best to innovate in end user applications, and I say so as someone working on publicly funded research software.

There is something to be said about the market's ability to make decentralized decisions and focus on satisfying people wants[1], so a centralized software economy is also not a good solution. The problem with markets here is that strong privacy and open source are, for the most part, positive externalities. As a user, the benefit you get from having strong privacy yourself is not usually noticeably high, nor that of having access to the source, specially for a non-technical user, yet society arguably benefits from both. Usually the answer to a problem of unaccounted externalities is government regulation, but in this case, large-enough-to-matter governments have been unanimously on the side of less privacy, rather than more (as is the case of the original article).

[1] Ideally, software should be designed so that it preserves privacy as much as possible while achieving its function, and is open source, and it provides all the million features and reasons people use something like Facebook, Snapchat, Youtube, etc. Just having privacy preserving software written for and by technophiles is not and can never be a complete solution.

Not quite the same, I can appreciate, but inroads are being made with the UK Digital efforts being mostly open source.

http://blog.quickpeople.co.uk/2013/05/17/the-uk-government-p...

For individual projects, it has already happened. BSD development, for example, was sponsored by DARPA in the 1980s. The German government supported the GnuPG project for a while. I am certain there are more examples - but you are right in that it has not been done systematically.

(Which does not invalidate your point one way or the other.)

Reply on news.ycombinator.com