So then shouldn't the link-local address be either MAC-derived or random, especially if you have multiple ifaces? The article's fe80::4 example seems weird. Mine is fe80::18cf:85a3:bc97:d117.
Note that even a MAC-derived link-local may not be unique, since the same MAC can be on both networks (e.g. with VLANs, or if you assign MACs to hosts instead of NICs).