factor-specific random-effects variance

I found this blog post: https://rpubs.com/bbolker/factorvar and it looks rather difficult to obtain factor-specific random effects (without the extra correlations). What would be wrong with the simpler approach:

dependent~fixedeffects+(0+dummy1a|randomvar)+ (0+dummy1b|randomvar)

where dummy1a and dummy1b are the same variable but coded inversely 0-1 and 1-0?


Regards,

Torleif