As pointed out in the previous answers, it seems quite clear that plus...operae is an argument of the verb poneremus. I found that some philologists corrected the text as follows: in agendo plus quam in scribendo operam poneremus (e.g. see here), which led me to misinterpret the syntax of this example (see the relevant comment by cnread, who alerted me of that confusion).
Let me address the second part of Joonas's answer/comment, which is more directly related to d/e's question, i.e. to the interesting title of his/her question: “gerund + genitive” vs. “gerund+accusative”. This is indeed a very intriguing issue of Latin syntax. Although the following examples are rare, they do indeed exist (typically, in Early Latin but even in Classical Latin as well: see Cicero's example below):
Lucis das tuendi copiam 'You give me the privilege of seeing the light.' (Plaut. Capt. 1008)
Nominandi istorum tibi erit magis quam edundi copia 'You will have the privilege of naming those things rather than of eating them.' (Plaut. Capt. 852)
Facultas agrorum suis latronibus condonandi 'the opportunity of bestowing lands on his fellows-bandits.' (Cic. Phil. 5. 6)
The genitive gerund of (some of) these examples has been analyzed as "epexegetic(al)", i.e., as "giving further precision to the expression -'opportunity of light, that is of seeing it'" (Palmer (1954/1988: 321-222); cf. also Woodcock (1959: 162-163), i.a.). Personally, I'm not fully convinced by this analysis since I'm afraid that it cannot be applied to all the examples that show this (rare) pattern. It seems more natural to me to analyze these genitive gerunds as having nominal behavior, this fact triggering the genitive case of their object. The double possibility (genitive vs. accusative object) shown in this contrast (e.g., cf. the infrequent constructional pattern facultas agrorum condonandi with the more frequent one facultas agros condonandi) reminds me a bit of the following one in English, where the object of the -ing nominalization can also be prepositional/genitive or not/accusative: e.g., cf. "John's destroying of the city was horrible" vs. "John's destroying the city would result in..." (e.g. see here and here). Clearly, this parallelism is only approximate since in Latin the construction facultas agros condonandi is by far more frequent than the attested example above from Cicero facultas agrorum condonandi (as is well-known, this gerund construction can in turn be compared with the very frequent gerundive construction facultas agrorum condonandorum; for related discussion, please see my answer of this question).
As for d_e's/Joonas's example ((in) scribendo epistularum), this specific constructional pattern does not seem to be attested. In contrast, as noted above, an example like facultas scribendi epistularum could exist. This example seems to involve a sort of blend of two possible "input" constructions: the gerund one (facultas scribendi epistulas) and the gerundive one (facultas scribendarum epistularum). If this proposal holds water, one could try to apply the cognitive linguistic notion of "input space" (see here) in order to account for the creation of (some of) these "blended" examples. Cf also the interesting example from Plautus: tui (feminine!) videndi copiast (Pl. Truc. 370).