The +/- and -/+ means you're heterozygous, you got a variant from one parent only.
Exactly
... it could make a smaller difference (or none) than the +/+ (homozygous).
Yes, though often even the homozygous version reported as +/+ does little or nothing at all. Unfortunately there's a lot of innocent SNPs which have been falsely accused of doing horrible things, such as by Yasko. But basically genes will usually behave in a dominant or recessive manner. If dominant, only one naughty allele (heterozygous) is needed for the SNP to affect gene function substantially. If recessive, then a heterozygous SNP on that gene will essentially have no effect.
Hence +/- can be a misleading label. It doesn't indicate a smaller impact, or intermediate risk between being -/- or +/+. It just shows that only one allele is present, and says nothing about whether it's relevant at all. So in addition to reading the original research to make sure a SNP hasn't been falsely accused, it's also usually necessary to read the research to see how much impact it has when +/- or +/+.
I'm not sure that the -/- is as innocent as it looks, because there are plenty of variants that are recessive, and the -'s make me wonder if the recessive variants are listed like that also.
-/- doesn't have anything to do with being recessive. It's used to (supposedly) indicate the lower-risk version of a SNP. However sometimes people use it for the more common version of a SNP, even if the less common version is protective and hence less risky.
There's a third possibility, that snp might be " turned off" so you might not have any worries from it, if I have that straight.
Genes can be regulated, with external factors in the body causing them to produce more or less of their protein. But SNPs basically never change. So a SNP is not regulated or turned off or on or up or down. People often extrapolate from the regulation of genes to suggest that something similar happens to SNPs, but they are just making stuff up to attempt to explain why research shows that a certain SNP (usually on the CBS gene) doesn't do what Yasko & Co says it does. Either the SNP does something, or it doesn't - it doesn't get "expressed" depending upon external factors. Only the gene can do that, in which case the specific SNPs are irrelevant, beyond the impact which they always have.