Early JWST observations have led to various unexpected discoveries, such as the high frequency of UV-bright/massive galaxies at z=9-16 and the existence of supermassive black holes at z=4-7 far exceeding the local SMBH mass-stellar mass relation. Several possible fixes to galaxy formation models have been proposed for the first issue, with the most widely entertained being a reduction of stellar feedback compared to what assumed by pre-existing models. On the other hand, the early existence of such SMBHs suggests an unexpected propensity of baryons of reaching extreme concentrations. While JWST had not yet reached L2, a substantial reduction of feedback was invoked to account for the formation of globular clusters, with their ubiquitous multiple stellar generations, which remains an intriguing puzzle in astrophysics. Direct evidence indicates that they formed in a series of bursts, and each burst did not prevent the occurrence of the following ones, as if there was no negative feedback on star formation. Moreover, second- generation stars typically exhibit different light-element abundances, but no sign of enhancement by supernova products. This, together with the lack of feedback, indicates that star formation took place before supernovae started to affect the ISM. It was then suggested that, above a critical mass, stars fail to produce supernova events, but rather sink into black holes without ejecting much energy and heavy metals. This would suppress star formation feedback for some ~10 million years, in practice leading to runaway star formation, analog to overcooling that in the absence of feedback would have turned most baryons into stars in the early Universe. If this is indeed how globular clusters formed, then a generic ~10 Myr delayed feedback would have important implications for star formation in general, and in particular for the very high redshift Universe, helping to account for the unexpected frequency of UV-bright/massive galaxies at z=9-16 and perhaps even helping the early formation of SMBHs.