Maybe there will be an algorithm for emulating stereoscopic views, but it cannot be the same.
The reason is quite simple. In a 2D video is not only missing the depth information (which is not sufficient for getting a stereoscopic video), but it is missing the hidden surfaces which would be visible from another point of view.
Everyone could think that the depth information could be extrapoled from the available information, and this is true. But, the missing information cannot be so accurate for having a good stereoscopic effect.
Apart from that, I've heard about a system which coudl extract accurate 3D models from 8 (eight!) cameras pointing on the same target. It is so accurate to emulate also clothes movements correctly. However, this is done processing 8 (eight!) 2D videos. How it could be possible to achieve the same result with only one 2D video?
The achivement of the wanted result depends essentially by the information availability, and in this case (IMHO) there is not enought information. Althought, it could be possible to try to emulate stereoscopic effect from a 2D video, but essentially it needs an hard work, long processing and the consequence is low quality result respect an original stereospic video.
I would like to remember that the 3D view is generated by our brain. The eyes can capture only 2D images, and our brain, processing the two images can generate a depth view of the seen objects.