In social science quantitative research context, to me I have to put in most of my effort to write "discussion" section of an article / thesis because a researcher needs to thoroughly understood all the previous sections i.e. research problem, research objective(s), research question(s), conceptual framework / research model, hypotheses, research method & actual results / findings. Some considerations when writing "discussion" section include:
1) discuss only based on actual result (if results are too lengthy, it is good to summarize the results for easier understanding).
2) explain why the hypotheses are supported / not supported, explain why the variance for some variables are high / low, explain why the strength of influence is high / low etc.
3) for those hypotheses not supported are they in line with your literature review / theoretical framework underpinned? (explain the same for variables' variance & strength of influence etc.)
4) if not what might be the reason / rationale / due to other factors etc - need to check & explain those reason / rationale / factors are justified with additional supported literature or not.
5) if justified with additional supported literature, need to justify why you didn't review those additional literature during the initial round of literature review.
6) for those supported & not supported hypotheses / new discoveries - what are the lesson learned & implication for society / community / knowledge contribution.
"Discussion" section should deserve the highest attention & longest writeup (if needed for reader to understand your discussion) because you can't refer to any help as you need to come up / explain on your own.
In many papers, I found that last paragraph of the discussion consists of strengths and limitations of the study. What do you think? It is mandatory there or, we can explain in methods section also...
Think up to individual researcher's preference so that the discussion flow is smooth & logical to the readers. Sometimes, where to park also dictate by some journals how they want to review / publish the article. For me the limitation, future research & conclusion usually I park under closure section so that closure section not too short only come with conclusion.
In scientific areas, based on articles i read, the last paragraph discuss about results summery and strength of the research. Though, different authors have different views