Primarily due to poor management of the project. The best way to showcase this is through something called "testground".
Testground was an initiative started by Protocol Labs to create a solution that would be used for testing of IPFS at all levels of the protocols, while also giving developers an SDK that can be used to write new testground tests. The reason this was needed is because the before the Testground project was started, the last few releases of go-ipfs (the reference implementation of IPFS) introduced new bugs, and regressions. Why? Primarily because of the poor planning, poor test coverage, and overall poor project management.
Tl;dr of that comment? Poor project management lead to a lot of wasted development time on Testground.
Irony of the whole situation aside, if this was an isolated incident it might not be so bad. But it's not an isolated incident, it happens all the time, and with every single project managed by Protocol Labs.
Interestingly enough this latest post from Axios caused me to come out of the woodworks with my own series of posts about how Protocol Labs has failed both IPFS and Filecoin. I briefly detail a few more of the problems here (https://bonedaddy.io/blog/ipfs/posts/failure-ipfs-filecoin-p...)
How so? I don't have much experience with IPFS, but I've heard great things about it