First, a minor nit, the Guardrails Approach isn't based on research from the 1990s, but from the 2000s.

Second, and far more to the point, there is a more recent, risk-based variation on the Guardrails Approach (https://www.kitces.com/blog/guyton-klinger-guardrails-retirement-income-rules-risk-based/) that does far better.

The main drawback of this new version is that it requires the help of an advisor who has the software tools needed to run Monte Carlo simulations to assess risk.

Finally, if one has enough resources to purchase a sufficient stream of dividends to cover one's retirement needs, that's certainly a good idea.