Millisecond Forums

Re. #1: Every time someone visits the launch page or refreshes it, a new (subject- or group-) number will be returned. Someone visiting the page but deciding *not* to participate at this moment will thus cause a "gap" such as those you observed. The only way to avoid this presently would be to explicitly assign your participants to a given condition (id) and pass that info in via URL query parameters.
Re. #2: Is impossible to answer with certainty. The participant might have mistakenly started the experiment twice. There have been some (rare) instances of double-executions in the past (e.g. https://www.millisecond.com/forums/Topic12943.aspx ), but this should not happen with the latest Inquisit web version.
Re. #3: This is impossible to answer without further details. The data files might offer some hints as to where exactly freezing (if any) occurred and what it may be related to. In general, such instances should be rare. However, there's always a chance you'll encounter some subset of systems in the wild with issues (running outdated and/or broken software or drivers, badly maintained) -- there's no way to avoid this, unfortunately. I should also note that -- sometimes -- people simply misunderstand what they are supposed to do in a given task (e.g. misread the instructions, mixed up the response keys, etc.) and may mistakenly believe the task "froze" when it actually did not. I'm not saying that's what happened here, just pointing it out as another option to consider.
Re. #4: For all intents and purposes this sounds like a data merging issue (I do not know how you merged the data and if all files had the identical format), whereas the value for the 'date' field ended up in the 'response' field. This may happen if you altered the data recording at some point (adding or removing columns), causing some files to have a different structure (fewer or more columns) than others.Hope this helps.