If there were as many ultra-luminous starbursts at z = 1 than at z = 5, the present sub-mm continuum surveys should have detected them (there might still be identification problems, however). A good way to identify them is to search for molecular lines, but the present instruments are not yet sensitive enough, when they are not helped by gravitational lenses. With more than an order of magnitude sensitivity that will be available in the next decade, it will become possible to detect normal star-forming galaxies at high redshift, and even protogalaxies at z > 5.