Keep in mind that every single living organism harbors lethal alleles. Most of these are recessive, and masked my heterozygosity. In populations, heterozygosity is a measure of overall diversity, and therefore ability to resist pressures and adapt, or a general measure of population health. This is why F1 crosses are said to have hybrid vigor. The heterozygosity created in an F1 creates hardy, fast growing plants. It will mask any harmful recessive alleles.
http://en.wikipedia.org/wiki/Heterozygote_advantage
When selfing a plant, you increase homozygosity by reinforcing dominant or recessive alleles. However, in a population of individuals - a given recessive allele for a gene in one individual is different from that same recessive allele in another individual - even though they have the same effect. For example, the autoflower is a homozygous recessive- however, one autoflower population may have one version of the allele (lets say the "lowryder" autoflower allele) while a different population has a slightly different sequence of DNA that codes for the same autoflower effect. (stitch's hungarian ditchweed or whatever he got his super auto allele from). This isn't bad for autoflower alleles to be identical by descent, but out of the thousands of genes in cannabis that will randomly drift towards homozygosity during your selfing project (what if it is cannabis cancer, or a gene that causes leaves to deform), this will inevitably lead to a depressed line.
Selfing increases the chances that BOTH recessive alleles come from ONE parent. This is a bad situation because this is where seed viability drops, plants make sterile pollen or sterile pistils, albino leaves, etc. Selfing is the quickest path to genetic deformities and inbreeding depression.
http://en.wikipedia.org/wiki/Inbreeding_depression
However, Highly inbred lines make great breeding tools. You can be sure that the phenotypes you see in the selfed line plant breed true - which will help when chasing those same traits in the f2 of a hybrid line. If you are using a plant for its lemon scent, but unbeknownst to you it is a heterozygote, (not a homozygote like a selfed line would produce), you will see no sign of lemon in your F1s, AND only half of the F1s will be beterozygous for the lemon trait, while half don't have the allele at all. However, IF in your inbred line you locked down blueberry scent, and all the resulting seeds show blueberry scent, you can be sure that your inbred line breeds true for that particular trait. If you were to cross this with something way different, for example thai, you could be sure that in the F2 you would have some plants that smelled like blueberry.