Open Source (Almost) Everything

Tom,

I agree with everything but the "core business" stuff and the MIT License.

The same principles apply to your core business - independent of what it is (not only GitHub).

By opensourcing all of your business, you will gain all the synergy you mentioned (i.e. libgit2) within your core.

Your competitors will benefit from it, but they won't be a challenge to you as long as they don't have the passion and insights from your team. If they can find a way to do better than you with that knowledge (expressed as software), then your position in the market is based on imperfect information (and therefore, monopolization power).

Although you may have everything to start a DVCS frontend, it is not an easy task which doesn't rely only on software: you need the ability to understand the infrastructure needed for the problem and maintain it at a reasonable speed - armies of proficient developers extending and fixing bugs, armies of sysdamins, coordinate them, decide adequate directions, etc.

Besides that, you need passion about what to do in order to keep kicking asses. Without it, it is not sustainable in the long run. Think about Launchpad, I think they add extra functionality to bzr than GitHub to git, yet people still sticks to GitHub. Why?

Also if you opensource your "core", people will need also customizations, so your competitors can become your clients where you provide developing to them - they may target other niche that you can't or your are not interesting.

This is where the license comes in: the best way of achieving this is by using AGPL3.

With AGPL3, you make sure that everything you gave it won't be restricted to others - as you haven't restricted anyone by opensourcing it.

You are not restricting others' freedoms, they can do whatever they want with the software. When someone closes the source, it's restricting others freedom, not his. What you can't do with AGPL is taking away others' freedom and right to know what they are using.

If someone improves it, then it will come back to you, and you will be able to improve from others as they improved from you.

The MIT license is good, but it doesn't close the loop and may have leaks, :D heheh!

A hug from Uruguay, Rodrigo

One simple rule of business: commoditize your complements.

Apple sells hardware and "gives away" all their media.

Content creators and developers are charged only enough (30% margin) to basically break even managing the system. This adds tremendous value to their hardware since they now have "an ecosystem" and not just some electronic widgets.

Aside: Apple have set prices for music but assumed app developers would always charge for their apps. When huge numbers of developers started giving away apps for free, Apple came out with iAds in an attempt to defray the cost of hosting free apps. They don't do anything for free.

So, the same applies to open source and hiring. Good engineers are hard to find. If you outsource your software and it becomes popular, good engineers will flock to it.

You've now got engineers pre-trained (for free) on an important component of your system and they are dead easy to identify, defraying the cost of recruiting.

Since open source is the commodity in this scenario, giving it away for free is the most prudent way of insuring widespread use and a good supply of engineers.

Edit: another idea is "give away your ideas, sell the system". Open source your non-core ideas to attract attention and sell your complete system. Although many marketers argue the reverse: give away your BEST ideas to prove your worth and build trust then sell them on the whole package.

Based on his post, yes, in his view, everything should be open source.

The product and data in this case is your "core" that the article is advocating to keep close to your chest while outsourcing the non-essential components that would attract interested parties to your work.

There are two schools of thought, in marketing, on giving away parts of your business, your ideas etc:

Give away your non-core ideas to attract people/resources/interest and then sell your complete system.

Give away your core ideas to "prove your worth" and build trust/authority/recognition and sell your complete system.

The second version is more powerful but it takes a leap of faith. Others could poach your work and use it against you but truly viable competitors are rare and people are lazy, the odds of this happening are often slim.