The "mistakes" question is percentage on the actual wordcount (mistakes per 1000 words), and not on what you fill in the questionnaire, so:

• If someone translated 5000 words and made 5 mistakes, it is 0.1%.
• If someone translated 4000 words and made 4 mistakes, it is still 0.1%

so there is no change there. The wordcount issue is not that easily fixed, we would have to go to a formula of $X/word and is something Utopian doesn't want from what I gather, as it gives the sense of a normal job and not an incentive.

As I've been saying all these months: we are all volunteers and we are not guaranteed to get a salary an upvote by @utopian-io. It might not sound right to some, but that's the truth, that's how the system worked since the beginning in all categories (and that was the purpose of utopian - to provide incentives and not payments). if it was sustainable, we wouldn't have to change the questionnaire and make it harder to get a better score, at all.