We address the issue whether ARPES measurements of the spectral function Aκ (ω) near the Fermi surface in the normal state of near optimally doped cuprates can distinguish between the marginal Fermi-liquid scenario and the spin-fluctuation scenario. We argue that the data for momenta near the Fermi surface are equally well described by both theories. but this agreement is nearly meaningless as in both cases one has to add to σ″(ω) a large constant of yet unknown origin. We show that the data can be well fitted by keeping only this constant term in the self-energy. To distinguish between the two scenarios, one has to analyze the data away from the Fermi surface, when the intrinsic piece in σ(ω) becomes dominant.