Better than offsetting one set of geoengineering (deforestation, ocean acidification, enhanced GHG emissions) with another equally radical one, it might be safer — from a risk management point of view — to try to not do any geoengineering at all. Certainly as long as our understanding of climate is just starting.
To borrow an analogy from medicine: preventing is better than curing. Patients are sometimes prescribed medicine where the side-effects are so bad that they're almost lethal by themselves, but only when the patient is surely terminally ill, so there is "nothing to lose". We do have a lot to lose on Earth.
Note that weather models are initial value problems and climate models solve boundary value problems, so mathematically speaking, the two are completely different.
Additionally, another major problem with geo-engineering is that once we've started these processes and essentially borrowed time to offset mitigation measures, we'll have put ourselves in a situation where these measures will need to be continued almost indefinitely, regardless of the risks of negative side effects of geo-engineering to begin with (ie: chemical intrusions, biological system degradation, geopolitical concerns). Present climate models suggest that if we were to engage in climate engineering and then remove these after several years, warming rates will come back with force (http://iopscience.iop.org/article/10.1088/1748-9326/4/4/045103).
Some things else to consider is the effect on different biomes of different sizes. The world literally operates differently for life of different sizes such as a microbe compared to ants, and an ant compared to us. An ant can fall off the empire state building and be just fine and us, well, not so much. Some insects are so small they swim through the air rather than fly. Thus we have to keep all sizes of life in our consideration. Seafoam may cause certain bacteria to flourish and kill others and we might not know for years. This effect might not be that big or it could cause dangerous algae blooms (this is a theoretical example not backed by any experiment). This plays into your LHC example. LHC operates on the smallest units we can manipulate. The earth is the largest unit we can currently manipulate and the rules are very different.
Going back to your original questions and worst-case scenarios we could cause issues with the earth's ozone layer, collapsing it and wiping out humanity, we might cause massive storms, we might kill off keystone species (some of which might exist on a microscopic level making it hard to detect until it's too late) causing great dyings releasing more GHG into the atmosphere.
The most realistic issue is that we don't apply geoengineering evenlyish (I add ish because optimal applications won't actually be even. Studies have shown that adopting certain areas, or engaging only certain areas with geoengineering, will make negative effects of climate change worse in the least modified areas. This means that is the USA implements geoengineering and Africa doesn't Africa will have a harder time. Conversely, studies have also indicated that well-implimented geoengineering can reduce inequality between nations which would drastically boost the quality of life and economy of the world's population as a whole.
Other than that the next most likely possibility is that geoengineering causes some large scale issues which will cause great dyings which will release more GHG into the atmosphere. Life is a giant cycle of carbon and other compounds and messing with that cycle causes levels of different chemicals and elements to change.
I see the argument that once we start geoengineering we can't stop. This is assuming GHG levels are maintained or increased (which is what we are going for) and suddenly removing a technology that lowers the temp by 0.5-2C would result in fast changes than would happen 'naturally'. However, this concern assumes that we only have 1-2 geoengineering projects to work with. A more realistic view is that we will use multiple methods in conjunction. Not only does this redundancy remove the concern that any single project can't be removed but it allows for some projects being better or safer in different circumstances. We will learn a lot, and it lowers the number of unintended side effects for any particular project.
I hope this helps. -Tristan