So we have learned that there are many ways we could interact with objects in VR, including different ways to select objects and to manipulate them. There isn't a best way or some kind of recipe for all. The way you design the interaction, new application, should really depends on your own specific scenario. How many objects are there? How cluttered the environment is? What is the size of the object you want the users to interact with, and how far is it? There are, however, some evaluation metrics which you can use to compare different methods. The first big category is task performance. How long does it take for user to actually complete the task? And how accurate are they in performing the tasks? The second category is usability related issues. How easy is it for the users to learn the interaction techniques, and how easy is it to use them? Is it likely to be something cognitively demanding? Or is it something quite intuitive? The third category, which is quite specific to VR, is about how natural and immersive the experience is. Does the interaction technique make the whole experience more or less immersive? Does they contribute or jeopardize plausibility illusion? Another factor which is important when it comes to object manipulation is expressiveness, which refers to the flexibility the user has with a given technique. Can I manipulate an object freely? Or does it make me feel that I'm being restricted to certain numbers of methods provided by the system? For instance in an interior design app, maybe the system allows me to move a sofa to a different position and orientation, but it does not allow me to scale or change a color of the sofa which might let me feeling frustrated and unsatisfied. Last but by no means least, the very important question, does your application takes care of user comfort? You will need to check if your application causes any kind of user discomfort including simulation sickness or fatigue. In particular, as users might be using their hands and head to select objects, then I have to keep their head or arms still and possibly in an awkward position for quite a long period of time, which could easily make them get tired. Or their performance might be very good in the beginning but later on as they're getting tired, their performance may drop. So make sure you also take that into consideration during your user testing. Different applications should have different focuses in terms of those evaluation metrics. In applications such as therapy and training, you might need to simulate the interaction to be as realistic as possible, so patients or users can really apply their skills learned in VR directly to the real world scenarios they want to be prepared for. So presence and naturalism will be the most important factor. If you develop a small game and you want everyone to play it, the interaction might not be very realistic but it has to be something intuitive so everyone could easily pick up. So here, ease of learning and user comfort are more important. If you are developing a tool for experts in a certain area, for instance, a data visualization tool for neuroscientists to visualize FMI data. It can be something to do have to spend some time to learn, but it has to be unsophisticated enough that once they have learned the interaction, it engages fairly low cognitive load so they can really use it as a tool to solve more complicated cognitively demanding tasks.