A "Lite" Engine for the replay system

First of all, hello everyone, this is my first post and I would like to apologize for my bad English. I'm French and really not good in other languages. :?

I search a bit on the forum about some idea that i have on the replay system but i don't find any sort of discussion about it.
The replay system look really awesome, some people ask about the size etc. But i think a big point with this Replay thing can be exploited.

Let's me try to explain:
As we know, technologies evolved every day and it's seem that Streaming or replay of game are gaining importance. And more than that, share this "new" type of media.
I think that the uber team as already think about a "Hub", a spot where you can share your replay etc. However, replay is "lock" into the game, you need the game to run it and that's a "problem" (personnal opinion).

I don't think that become a reallity but imagine a format that can be read by an external software. First thought: Useless, i prefer open the game and interact with the replay. But if we can use a minimal version of the engine to just look at the replay that can be awesome. (When i say look at the replay, it's with control etc, not a video, the entire 3D game experience but just as a spectator)

It's certainly really hard to do but just think about a HTML 5 or web format of the engine where you can play the replay file (local or hosted somewhere).
If this type of thing is possible we can imagine a Portable "Replay viewer" on mobile device for exemple.
Play with your friend and take the replay on your smartphone to talk about strategy, mistake and other things with the entire replay in your hand.
The possibility about this "Replay" or "View" system are infinite. I dream of a live Web viewer, a spot like Tw*tch but you can navigate in real time in the game without any other things that a browser. Or just an external software where you can edit your replay and post a video of the most epic moment etc.
The portability things is really interesting, and Android or/and iOS apps would be awesome.

So what do you think about that, an "engine" lite and portable + Web for replay files.
Some other idea about the format or different utility of the replay system ?
Probably not happening but imagine an Open lite engine, moddable etc. Community would create Soft on any plateform to use, view, stream replay and that can be big point for the community. For the E-Sport too, that can be an easier way to stream: Not using a external software but use the Web Viewer to Steam the Replay file in almost direct. With a moddable "Engine" only for replay we can also imagine that the Streamer (the guy which comment the game) can talk direcly trought this file etc.

What's your thought on that ?

P.S: Really sorry for my English >< i hope that i'm understandable ^^'

I dunno about what kind of technical concept would be best, but having a html5-replay viewer freely available would be quite awesome indeed. But I doubt they have time to develop such a thing at the moment.

I'm not sure if that would be possible.

Using the original client to jump in adhoc into a running game as observer? That should work theoretically, at least if i understood the implementation right.

But trying to write a fully featured webplayer? I don't think that would be possible. Well maybe possible, but it would mean very much effort. After all the replay is minimal, while it does contain sufficient data to show a accurate replay, it still relies on the game data in many ways. Take just the world generation for example, the web client would need to be able to parse the replay in such way, that it can display all relevant details of the terrain. This however would mean that the web client would need to implement large parts of the procedural planet generator which also generates a lot of overhead, to be specific all the stuff which just serves cosmetical, but no functional purpose.

This creates a giant game state which would need to be handled by the web client. This is no problem for the original client, it has several culling methods implemented which can eliminate the need to compute the currently invisible parts of the simulation. All these features must be implemented in the web client in order to enable it to handle the display of the simulation.

I don't say that it would be impossible, but writing that web client would mean almost just as much effort as developing the original client, that is if it was possible at all, because after all WebGL or Javascript in common do not offer the same performance as a native client where all OpenGL features are accessible.

This actually seems like it would be feasible. A client alone would only be able to view games and not play them since you'd need the server for that.

I don't see how you could make the engine "lite". What web technology do you think could be (re-)used?

A complete port of the PA engine to web technology (so, almost the entire game, written in inefficient code, with the corresponding download size and performance needs)? VRML? X3D? Every angle of the game saved as x264 / webm video file?

Actually I think this will be fairly doable. If someone whipped up a sphere with some dots it would probably be fairly easy to interpret the data.