Dragonfly: An optical telescope built from an array of off-the-shelf Canon lens

First is weather. We can't see through clouds. Most new astronomy is about sources too faint to have been analyzed a hundred years ago, and even clouds that are barely visible to the human eye will drown those out.

Second is various engineering difficulties resulting from differential temperatures in the air in close proximity to the telescope dome, defects in the mirror surface, and limitations to the optical design (you're projecting a spherical globe onto a flat surface).

Third is 'atmospheric seeing' - high-order distortions caused by thermal patterns in the air which change significantly on a tens of milliseconds timescale, ultimately leading to a gaussian blur of the light in long exposures. The lower your altitude, and the more disturbed the airmass, and the more humid this is, the worse this is.

Fourth is sky glow - light pollution from nearby upwards facing lightbulbs, from the full moon, and from the sun at twilight & in the daytime

Fifth is the diffraction limit. A perfectly engineered, spherical-cow-world telescope with a perfect sensor has fundamental optical limits to the resolution it can observe, and optical resolution in arc seconds scales with wavelength / aperture.

Sixth is bright-source confusion and the limitations of your background field. It's very difficult with CCD & CMOS sensors (and even with spherical-cow sensors, the optics present limitations) to image a faint thing next to a bright thing. This is why we have fewer galaxies mapped on the other side of the Milky Way,, and why it can be very difficult to pick up, say, a nebula right next to a bright nearby star

Seventh is light-gathering ability, thermal noise, and readout noise. If you're trying to capture a photon every second, it's going to be very difficult if your CCD is absorbing thousands of photons per second thermally from the surrounding blackbody radiation and the readout circuitry.

Eighth is differential focus. To make matters more complicated, optical resolution is not 'fixed' because focus is not identical in different parts of the iamger; Typically telescopes are optimized for nominal focus at the center of their field, but get a few arc-minutes off of the center and optical resolution goes down. Get a few degrees off and it can go down to un-usability. There are characteristic abberations that crop up, and every optical design that aims for wide fields is a compromise between these abberations.

Ninth is atmospheric windows. Atmosphere absorbs hard UV. And portions of infrared. And portions of radio. To get a full spectrograph of a source, to detect the exotic portions of the EM spectrum that we don't really deal with frequently, you can't do it through atmosphere.

Generally speaking, it's relatively easy with on Earth for professional observatories to reach a point where atmospheric seeing limits your observations more than diffraction or readout noise or field distortions or sky glow or ambient light. It's not easy to defeat bright-source confusion with a larger and larger telescope. Many astronomers have had to content themselves with knowing little about the sky right next to bright sources like nearby stars. The telescope in the article tries to probe this known unknown with numerous small low-res cameras.

Space observatories provide us a small amount (10x?) better surveys because of no sky glow, daytime observations, no weather, etc. They eliminate atmospheric windows and simplify some engineering issues (while complicating others).

Part of the big remaining purpose of space observatories, the thing it's very difficult to do on the ground (we've tried!) is to defeat the atmospheric seeing limit and allow us to use very large telescopes which are relatively simply designed. Light-gathering ability from a source scales with aperture^2, and light-concentrating ability scales with aperture^2, so ideally sensitivity to sources should scale with aperture^4. It rarely does on the ground, because we have to put up with atmospheric seeing. The technologies we've used on the ground to fight atmospheric seeing are extremely limiting, expensive, complex, the subject of an inane number of PhD theses, and only suitable for very small fields.

This goal of survey astronomy is at cross purposes to the telescopes in the article, which aim to get diffuse low resolution impressions of the light near bright objects, defeating problem number 6; They can do this with relatively short exposures over hundreds of sensors, so that none of the electron wells in the sensors ever saturate from being full of too much light and spill over into their neighboring electron wells