Matthew Ström, design lead at Bit.ly, is also picking up the torch and warning about in-product messaging (aka what a lot of people think about when they think about user onboarding). He correctly calls out the potential opportunity cost of getting locked-in to a bad design when using in-product messaging to fix things, but takes a different tack about it than I did. His point is less about calcifying a bad design with hard-to-change tooltips, and more about obscuring the fact that it's a bad design decision in the first place.
Frequently, these tools are used by marketing departments to pursue marketing goals. When one of those goals is "conversion" and your product/development team is not also given that same goal, your marketing team can use these tools to shore up problems without the development team ever getting involved. This seems great, except for the fact that it's a band-aid fix. Your product team is now unaware of the problem in the first place, and can't resolve it. Because they never fix it, the tool-tip bandaids become a permanent fixture and further features now have to have them to not stand out. Your product is permanently more complex and the underlying usability issues remain and are perpetuated forward.
I heartily recommend reading the article for yourself. He goes into some wonderful real-world examples of where this kind of thinking leads to, some that I've experienced and continue to be frustrated about, others that are new to me.