frederik, could you tell us your point of view about this?

Randomization should only be done once. So neither going back and forth, or changing the language, or saving and resuming, or doing whatever should change the original order.

Furthermore, it is super important that each variable is randomized independently. So for example if I have three randomizations with two groups (A, B), there has to be a "coin toss" for every randomization (resulting in either A-A-A, A-A-B, A-B-A, B-A-A. A-B-B, B-A-B, B-B-A, or B-B-B ). It would not be good at all if all randomizations would always have the same value (A-A-A or B-B-B only).

I know it from other survey software, that randomization for each variable is done and saved in a variable (!) on the start page of a survey.

thank you. Yes we're already using seed for the randomization, olle implemented it for LS3.

@olle: can you tell us if this part of the randomization bug depend on your seed system?
If yes, just assign it to dominik and explain him how to fix when you have some time.

Confirmed, will fix.

What is the status of this bug. Is the seed changing too?
Perhaps the plugin https://gitlab.com/SondagesPro/getQuestionOrder should be implemented to record display order.
We might see a lot more issues around randomized display order. Hard to catch with the eye alone.

For any move into the scientific community, a correct documented display order is crucial.
https://www.limesurvey.org/forum/future-features/118431-viewing-order-data-for-randomized-surveys

Status is that I forgot about it.

@cdorin, can't reproduce. Double check?

Hmm, I also cannot reproduce it any longer (3.17.3)

Cannot reproduce anymore.

Version 3.17.4+190529 released