1/ Yesterday, the day the GitHub acquisition was announced, @fredwilson had a good post on decentralization and data portability: avc.com/2018/06/why-de… 
2/ The post also mentioned that it's possible to get data out of GitHub. It's not trivial, but it's doable.
Here's a graph of projects imported into one of GitHub's competitors, GitLab, over the last 90 days. There was a huge spike when the GitHub acquisition became public news.
Here's a graph of projects imported into one of GitHub's competitors, GitLab, over the last 90 days. There was a huge spike when the GitHub acquisition became public news.
3/ I have a different take than Fred on lock-in: I think decentralization & low switching costs don't necessarily go hand in hand. You can have one, both, or neither of those in a product.
4/ One can have a centralized platform that makes exporting your data super simple (e.g. @Pinboard). One can also have a decentralized platform that stores your data in a proprietary format and doesn't have an easy export feature, or that charges a heavy fee for exporting.
5/ IMO switching costs are about openness & common data standards, not centralization. If you can't export data, that's useless. If you can export data in a proprietary format that others can't import, that's also useless. Doesn't matter if the underlying tool is (de)centralized.
6/ Finally, switching costs can be an effective moat (w/high market share), but they're also a lame moat. If the only way you can keep customers is by making it hard for them to leave, then you have a crappy product. Build a moat around creating more value instead.
