The notion of what constitutes a typical planetary system has undergone some serious revision in the past twenty years. Our own solar system, once seen as a timeless and almost mechanical entity, is now known to be on the margins of chaos. Long term modeling of its dynamical evolution suggests that orbits of an inner world like Mercury may evolve over billions of years into something unstable, eventually unleashing a dramatic revision of other planetary orbits, including those of the Earth and Mars.
Exoplanetary systems show signs of even more dynamical rearrangement, from migrated giant planets to violent dynamical 'cooling', in which young worlds can be ejected wholeheartedly into interstellar space. These rogue worlds, or 'nomads' really do exist, gravitational microlensing surveys have picked up the tell-tale brightening of distant stars that can only be produced by a population of small dark planetary bodies drifting around in the Galaxy. There are twice as many of these Jupiter-sized or larger objects than there are normal stars in the Milky Way. That is an awful lot of planets alone in the dark.
Except they don't necessarily stay alone. A rather intriguing new paper by Perets & Kouwenhoven investigates the likelihood that planets ejected into interstellar space by their parent systems (again, as part of planet-planet gravitational interactions during early stages of system formation) might be captured later on by other stars, meekly taking up a place at the edge of the orbital table in these systems. This process is helped by the fact that most, if not all, stars tend to form in associations or clusters. Our own Sun is now generally thought to have formed among numerous sister stars, now long dispersed by their dynamical to-ings and fro-ings, and by the effects of Galactic scale gravitational tides as everything is conveyed around the galactic center. Measurements of radioisotope remains in our system support this hypothesis, by indicating the past presence of massive stars that once exploded as supernovae in our near vicinity some 5 billion years ago.
Allowing for this type of transient stellar clustering Perets & Kouwenhoven calculate that the odds of ejected, nomadic, planets being captured into large, long orbits on the outskirts of entirely different star systems are actually moderately good. If a lot of planets are ejected from their original systems then re-capture probabilities can be as high as 3 to 5%, which may not sound like much, but if the stellar cluster has 1000 members that's a decent number, and if we extrapolate to the Galaxy as a whole that would mean a lot of 're-purposed' planets lurking around the outskirts of stars - including binaries and stars with their own pre-existing planetary systems.
This can help explain some of the current direct imaging detections of planet-scale bodies in extremely large orbits (100 astronomical units, and further, from the star), which are otherwise quite a challenge to account for in standard planet formation models. It also makes an intriguing prediction that nomadic planets can be captured by stellar remnants - like white dwarfs, and particularly black holes, whose high mass increases the odds of planet capture.
Given the relatively weak binding of these nomads to their adoptive stars it also seems possible (in my opinion) that some of them might even end up being ejected yet again, drifting through interstellar space, and then once more taking up with another system along the way - at least during early times while the stellar grouping is denser.
So we may need to further loosen our conceptual picture of stars and planets as isolated islands. Planet formation is messy; youthful (and old) planetary systems can be touched by chaos, and stars - even black holes - may captain the ultimate in leaky boats with an unpredictable and motley press-ganged crew.