In the past 5 years, the necessity as well as the opportunities to attack the problem of starbursts using a multi-wavelength approach have become manifest. No single band suffices, and the full EM spectrum from radio to gamma-rays is now enthusiastically embraced in starburst research. Some examples: (a) stellar ages and abundances are best deduced from UV-optical-nearIR observations; (b) the best estimator is L(UV) + L(IR), meaning that different instruments are always necessary; (c) a long-wavelength baseline is essential to overcome distortions by extinction of statistical samples and of physical inferences from any given band; (d) starburst regions can be opaque even at mid-IR wavelengths; radio/mm observations are needed for the youngest (~ 2 Myr) embedded sources; (e) mid-IR photometry and spectroscopy, now just coming into their own with the Spitzer Space Telescope, show great promise as dust/gas tracers within starbursts.
Understanding the physical coupling mechanisms between wavelength domains is essential: (a) there has recently been good progress in modeling the UV through IR spectral energy distributions of starbursts taking all three major components (stars, gas, dust) into account, but this remains a key area for additional effort; (b) the long-recognized relation between radio continuum and far-IR dust emission in star-forming systems is sometimes said to be the best correlation known in extragalactic astronomy, yet we do not fully understand its origins or implications for the star formation process. A less well established radio/X-ray correlation in young systems is also important to understand.
The fastest increments in observational insight into starbursts are currently coming from Spitzer and GALEX (IR and UV). Probably the fastest increments for the coming decade will be from ALMA (mm-wave).
Although this conference emphasized observations, we cannot forget that theoretical and computational astrophysics have to be part of a "pan"-discipline approach to starbursts.