Deblocking is automatically scaled by quantizer, so there's generally little need to adjust it by much. The specified values aren't actually the deblocker's strength, they're just an offset (or 'bias') that gets applied. I believe the offset works by adjusting the quantizer value used for computing the deblock strength/threshold. In other words, for negative values, the deblocker pretends that the clip was encoded at higher quality levels, and hence, has less strength.

As mentioned, deblocking is scaled by quantizer, so for high enough quality settings (somewhere like QP<=15), the deblocker is effectively disabled (which it probably is for this clip), so specifying negative values actually would have zero net effect for this particular case. However, huge offsets generally are undesirable, and probably signify a lack of understanding of how the deblocker works.