Who decides the paper acceptance rate at a workshop organized as part of a conference: the workshop organizers or the conference organizers?
This can depend on the conference. My experience (CS, various ACM conferences) is that the workshop committee is responsible for its own papers and agenda.
But the workshop has a limited time frame, possibly determined by the conference committee (time * space = money). This can have an indirect affect on the rate of acceptance since it implies a limited number of papers that can be accepted.
So, it isn't a decision that, say, 10% of the papers submitted will (likely) be accepted, but that there is only room for 10 papers, no matter how many are submitted. Some others might get shunted elsewhere, such as to a poster session.
But I assume that isn't universal and can hypothesize about a number of different possible arrangements.
To supplement some of the other answers, I think in general the process is as follows:
- The conference issues a 'call-for-proposals' for workshops;
- Workshop organisers pitch their idea, often with some indication of the intended scale (half-day/full-day/etc, expected number of participants);
- The conference accepts the proposal and informs the authors of the gross parameters for their workshop ("we are assigning you a room of X capacity of Y hours");
- The workshop organisers decide how to structure their timetable (1 hour = 4x15 minute slots? 3x20 minute slots? 1x30 minute + 2x15 minute?) and advertise for submissions;
- Once submissions close the organisers take stock of what has been submitted and try to construct a detailed timetable, perhaps tweaking their structure to suit submissions.
Of course, there are deviations from this structure: for example, sometimes the conference will wait and see how many submissions different workshops get before making room assignments. Also, in long-running conferences certain parameters (e.g. slot duration) may effectively be known in advance.
Thus the acceptance rate is influenced by three groups: the conference organisers, who control the overall resources available to the workshop; the workshop organisers, who choose how to assign those resources; and submitters, who control demand for resources.
Some additional thoughts:
- In contrast to the associated conferences, workshops usually don't have a strong incentive to keep the acceptance rates low. In fact, there's an incentive to accept as many acceptance-worthy submissions* as possible, since "number of attendees" is a relevant metric if the workshop is to be continued in the future.
- Many workshops struggle to get a decent number of acceptable submissions. It's not uncommon that workshops are reduced in length, joined with other workshops, or filled up with invited talks.
- In case that a workshop indeed gets more acceptable submissions that it can accommodate, there's the possibility to make the presentation slots shorter.
(*) Of course, papers still need to reach a certain quality threshold, since poor presentations and papers would reflect poorly on the workshop.
10% acceptance rate? That sounds more like a highly competitive funding agency rate. Yes, there are successful meetings that advertise that rate; but that's because they get many, many crappy submissions; I know a couple. And they're close to some lucrative industry.
In real life the organizers of a bona fide meeting have one big worry: getting enough people to show up so that the budget works out. Typically the organizers are personally responsible for deficits, or if they can manage it, their institution. It's also why conferences end up with tiers of industry sponsors (though depending on the commercial value of your field). The acceptance rate is derived from that. It's also why the acceptance mail state "at least one author has to register at full price". You pay the full price so that your student(s) get to go and you can stay home and get some work done.
It all boils down to: how many bodies do we need to break even? Acceptance rate derives from this. Yes, attendees at the opening ceremony get to hear that the acceptance rate had gone down from 51.2% to 49.1%. But that simply reflects the success of publicity efforts.
To be fair there are meetings (like NeurIPS) that have huge attendance and low acceptance rates (~20%). But that probably has more to do with the hotness of the field.