The universe is so unimaginably large that parts of it are literally impossible to see, because their light is too distant to reach us even after 14 billion years. Now we might know just how big the cosmos really is. Turns out everything we can see is less than half a percent of the entire universe.

So just how far can we see into the visible universe? Since the universe is about 14 billion years old, it seems obvious that we can only see within the nearest 14 billion light-years. But that isn't quite right, as cosmic expansion has expanded the distance between us and the most distant cosmic objects to as much as 45 billion light-years. That means that there's just about 90 billion light-years worth of visible universe that we can see.

But even that can't compare with the full size of the universe. Of course, if we can't see part of the universe, then we can't actually know anything about it, because no other information will have had time to reach us either. But the trick is that the structure of the universe we are able to see can reveal its overall size. Basically, the universe can have one of three structures: a closed shape like a sphere, a flat shape, or a completely open one. Either of the last two would mean the universe is infinite, but a closed spherical shape would mean the universe has a definite volume.

There are some subtle ways to measure the curvature of the universe. Very distant objects will be noticeably affected by the curvature of the universe - if it's closed, then the object will seem bigger than it should; if it's flat, the object will appear to be the right size; and if it's completely open, the object will look smaller than it should. A lot of different measurements have been made, but they give tons of different answers about the universe's size and curvature.

The trick is to combine all the conflicting data in some sort of useful way. Researchers have now used the Bayesian model, a form of statistical modeling that examines how likely given models are to be correct based on the available data. This form of modeling is far more powerful than those used before and has placed some very tight constraints on just how big the universe can be.

Here's what they found - in all likelihood, the universe is flat, which means it has an infinite size. But if it *is* a closed sphere, then we have a lower limit for its size, which is about 250 times the size of our visible universe. Obviously, 250 times to infinite is a pretty big range, but it's by far the tightest constraints ever put forward, and having an actual lower limit for the universe's size could prove hugely useful in other areas of cosmology.

For more on how the researcher came up with this, check out their original paper or this post on Technology Review.