Tightness is a value that indicates how well an Object Point's marks agree.
The tightness of a point relates to how well its defining light rays intersect. The better and closer they intersect in 3D space, the smaller the tightness value.
Take for example a point marked on only two photographs. Once the orientation and 3D solution is complete, the light rays from the two camera locations, through the marks, and out to the 3D point in space can be computed. If everything is perfect (perfect camera orientation, camera calibration, and mark positions) then the two light rays will intersect in 3D space perfectly - and this is the position of the computed 3D point. In this case the tightness value for this point would be zero.
In real-world situations neither the markings, nor the calibration, nor the orientation are perfect and these two light rays will not intersect - they will graze by each other - there will be a small gap between them (at the point they are closest to each other). The size of that gap is the tightness value. The larger the tightness, the larger the gap between the light rays.
When a point is marked on more than two photos, the tightness value shown is the largest gap between all light rays for that 3D point pair wise.
Tightness can be displayed in the Point Table as either an absolute value in scaled project units, or as a percentage (percent of the size of the project (the maximum diagonal extents of the 3D data in the project)). A tightness of 1.0% would mean that the light rays for that point miss each other in 3D space, and the size of the gap at the closest approach is 1% of the project extents. Note the 'Tightness as a percentage' column in the Point Table is not displayed by default but can be easily added.
In a high accuracy project the tightness values should generally be under 0.01%. A tightness of 0.01% means the gap between the light rays is one ten-thousandths of the largest project extent (0.01 / 100 = 0.0001 = 1 / 10,000).
In a good quality project with many manually marked points (not subpixel targets), the tightness values can reach 0.1% which translates into one one-thousandths of the project size.
Large tightness values can be due to imprecise marking or poor camera calibration.
Comparing tightness to marking residuals (see Residual and Marking Residual) - residuals relate to the error of an individual mark (how far is the 2D mark on the photo from the projection of the 3D point on that photo) and are measured in pixels, whereas tightness relates to the error of the intersecting light rays in space, as a 3D distance measure, and is displayed as a percentage of the project extents or as an absolute value in project units. See also Precision (which is another estimate of error that is computed by the processing algorithm).