Very odd.
Most defenses that give up a lot of big plays are generally terrible at everything but that does not define Ohio State's 2018 defense at all. They were very much feast-or-famine in that they were either shutting opponents down or giving up huge plays.
So, this is actually a little inaccurate, give or take quibbling about "a lot." It often reflects strategy more so, give or take execution/quality.
Using the Bill C stats, the two main offensive and defensive numbers mentioned are that IsoPPP (measures how explosive the explosive plays are) and success rate, which basically says, how good is this team at stringing together chain-moving plays and marching the field.
A bend-don't-break defense, the kind we oft lament, says we'll give up some success rate for good explosiveness defense (I've seen one coach argue big play prevention/production is a top indicator of winning, which in this context is debatable). Since this is all a continuum, it means there are some defenses that will try to be more aggressive. That means trying to get opponents off the field faster, with the risk of getting beat on a big play.
I've tried to figure out a name for this. Don't-bend, sometimes-break feels weird, and high-risk, high-reward fails because people are poor at risk assessment.
Examples of this when it works are Clemson at times in this run and old MSU teams. While those are both good squads, it's a low margin for error defense. They tend to be quite good or quite bad. Bend don't break has a higher floor, lower ceiling, and if you have good talent, teaching and technique, and strategy is fine.
Schematically this can manifest in concrete ways. If I put an extra man in the box to make short runs harder, I have one less on the last line of defense. If my corners play tight to take away hitches, the defender has a better chance to get behind them.
So let's look at OSU. Last year, this was their defensive rank:
36th in success rate (drops to 57th with another adjustment about down and distance)
115th in explosiveness
If the former was better, the latter would maybe be OK. What's more interesting is when we split up run and pass.
Vs run
41st in marginal efficiency (better version of success rate)
120th in marginal explosiveness
Vs. pass
52nd in Marginal efficiency
61 in explosiveness
So the pass D was kinda flat mediocre at it all, run D was try not to bend, break badly
Previous years
2017
Overall success rate 14
overall explosiveness 11
Run success rate 7
Run explosiveness 7
Pass success rate 35
Pass explosiveness 8
2016Overall success rate 14overall explosiveness 53Run success rate 9Run explosiveness 66Pass success rate 24Pass explosiveness 30
2015Overall success rate 7overall explosiveness 33Run success rate 25Run explosiveness 32Pass success rate 1Pass explosiveness 66
2014
Overall success rate 24overall explosiveness 29This doesn't look at some of the overall numbers, which get adjusted for competition. That raised the OSU D to 29th last year. I sometimes shy from adjusted numbers because they tend to go against feelings and confuse