so what? it shrinks, it grows. stop and think about what you're doing.
you're not trying to set the fence 5.6735" from the blade.
you're trying to have zero distance difference from one end to the other end of the miter slot.
it does not matter if the fence is 5.6735" from the blade or 10.00001" from the blade - is the blade / fence parallel to the miter slot is the question / aim / goal.
if you slide the dial indicator down the fence and it jumps half-way around the dial, that could indicate some attention is required....
How close is close enough?
that is for the individual to decide based on their own needs and degree of workmanship.
however, it does make a difference. the typical thin kerf blade = 3/32 = 0.09375"
if you're happy with +/- 1/64" (0.015625") blade alignment, the saw has to chew out an additional 2x(1/64) - 30+% wider kerf/more wood.
if one makes the effort to align the blade and the fence, I - at least - notice the difference in push force, work piece steering, and eventually end accuracy on furniture size pieces of wood. slicing 1/4" slats off 5/4 rock maple - also noticeable. I'm very prone to cutting stuff over size, then fine cutting it to the final dimension - meaning I'm often taking off less that half a kerf width - so a gliding-no-effort pass thru the saw makes a difference in my world.
I like to keep the saw heel within .002" and the fence within 0.005"
this can be done with a stick or a combi-square, or a dozen other "tools" - or a dial indicator. the degree of bending / stooping / peering / light gapping / effort / experience varies.
I have to chuckle at the dead batteries theory. my digital caliper has an auto shut off - the battery is still going strong at 4 years and counting. the Wexley digital protractor at three years and counting - so it's something I personally can live with - especially as I have a spare battery in the drawer - which by the time I need it, will likely be dead.
Read the whole post. Almost the entire thing discusses blade runout (wobble) and the non-need for ultra accuracy in measuring it. Too many sloppy factors go into building a common woodworking table saw. Spindles are not straight, bearings are loose, and blade washers are ugly stamped steel affairs. Unless you super accurize the rotating parts of your unit (which is not easy nor cost effective), you WILL have runout on your blade. This runout is why setting up the blade to the miter slot is accurate ONLY if the measurement is taken with the blade in the exact same spot every time. Rotate the blade as little as a quarter turn and your indicator will read a different number. I can align the blade and slot with a metal rule well accurately enough that you would never know the difference performance wise.
I do not crosscut on a table saw because it is almost always more difficult and less accurate to move the work than the tool... and it is almost impossible for a TS cut to be more accurate than a tool with the blade on top where you can align your cut by eye much more precisely. Therefore, at least at this time, I couldn't care less where my miter slot is in relation to the blade.
There are also much faster and easier ways to competently align the fence parallel to the blade without taking complex measurements. And again, none of these measurements are used on the Hubble telescope. There is no need for the blade to be a half a thou to the fence when, again, rotate the blade a quarter turn and lose that half. I will put my table saw performance up against any mass production built saw out there. I have for many years, and still do, make very nice examples of woodcraft. I do not struggle with my tools, nor do they let me down in the accuracy department. Any mistake made is by me, not my tools.
By the way, I'll bet most of you guys and girls lay out the measurements on your boards using a tape measure and pencil. There's your 64th (if not a click more) of an inch loss of accuracy folks.