Let $bo$ be the connective cover of the real $K$-theory spectrum $KO$. This is a ring spectrum, and so one can look at its Adams spectral sequence. Mahowald does this in "[$bo$-resolutions](http://dx.doi.org/10.2140/pjm.1981.92.365)", and starts working from the $E_1$-page instead of passing to the $E_2$-page, presumably because $bo$ is not a flat ring spectrum. Why is this the case? Mahowald explicitly computes the cooperation algebra $\pi_*(bo \otimes bo)$ to show a lack of flatness, but even before this he seems to understand that $bo$ isn't flat: in the introduction of the paper, he phrases everything in terms of the $E_1$-page. Is there any intuition for why $bo$ should not be flat beyond looking at the cooperations?