Most machines at our level will flex under a cutting load, follow the components between the workpiece back through the chuck and on to the tool and there will be a number of areas where this flex will manifest itself, and one of the biggest culprits is the shank of the tool itself.

It's a sensible policy to ensure that the amount of unsupported tool should be kept to a minimum. This unsupported length is called the 'Overhang'. With a regular RH knife tool the overhang need be no more than is necessary to allow the toolpost to clear the workpiece, so it can be very minimal. For most applications, so long as the cutting edge position is right, the tool needs to overhang by just a little more than the depth of cut. So arguably it might be around 1/4" – 1/2" on a small hobby lathe, but with reality kicking in it's more likely to be the top end of that, if only so that we can see the cutting action without peering over the toolpost too much. With this in mind then the tool is in a good place, the opportunity to be deflected down by the cutting forces is minimal.

With a boring bar, depending on the depth of the bore to be machined, we have to have a tool that overhangs much more, so a two inch deep hole would have a tool overhanging by over four times more than our knife tool. As such it's now in a less good place, without that support the boring tool shank can bend downwards under the cutting forces, and it also exerts more force on the toolpost because of the 'lever principle', this puts the cutting edge below centre, resulting in reduced side clearance, a tendency to snatch into the cut with compromised relative cutting angles effecting the shear plane etc. So the idea is that we predict that degree of deflection and 'dial it in' to the tool height, so that when it deflects it comes down to the desired height. So this is why dialling in a precise setting really is academic, as unless the materials, tool, overhang, speed, feed, and depth of cut are always the same, it's just a stake in the ground as a starting point, and for that you'd might as well set the tool on centre, and take a quarter turn on the tool holder height adjuster (or add a shim under a fixed post), or whatever, to act as that starting point, then take it from there by judging the cutting results to improve the cutting action.

To minimise the variations:

Always use the largest boring bar that will fit into the pilot hole, and move up to bigger ones when you can.

Use the minimum overhang.

Try and arrange the cutting tip over the saddle by retracting the cross slide

OK, so we're now cutting and we are looking to see what can be improved. A machine will always make it apparent that it's not happy. This shows itself as noise, vibration, an excess of current draw, and swarf with properties that can be judged.

Noise:

It's a machine, it make noise, but listen to the nature of the noise generated by the cutting action. A happy tool will sound of nothing more than a sort of hiss, high frequency sound with perhaps the odd 'snick' sound as chips separate. An unhappy tool will growl, lower frequency and a detectable rubbing sound.

Add the machine's sound to that, a motor labouring, bearings perhaps groaning, maybe even drive belts adding to the orchestra of audio output.

Vibration:

Probably the easiest to detect, again sound frequency relative, it can vary from lower frequency rasping to high frequency squeals all detectable through the fabric of the machine itself. 

Current draw:

Years ago it was common to fit ammeters to machines, these tend to be found on higher end stuff these days, but it was a way or a machinist to measure the strain being put on the motor. Modern machines may have variable torque which will compensate for any speed reduction under a cut, but this is often still detectable as sound variation between free running and loaded conditions.

Swarf:

A shiny surface on the cut face of the swarf indicates a sharp tool, parallel lines along the length show faults on the cutting edge, but most importantly lines running across the width of the swarf are indicating chatter or vibration. The other side of the swarf is usually matt (flat) in nature, as it consists of the previous cut surface of the material crunched up and separated as a normal part of the cutting action, expect this to be as even as possible.

As a side issue, you may often find that when using TC tools they can be at their happiest when taking a full capacity cut, and less so on shallow cuts. This is because TC can not be made as sharp as HSS, and if the depth of cut is not sufficient what is in effect a radius on the cutting edge will want to try and ride over the top of the material, rather than 'getting under' the cut. When they rub like this a poor finish and dimensional variance occurs. This is why it's common to say that it's wise to leave sufficient material for the tool to bite into for a finishing cut when using TC, rather than taking it down to near size and attempting a shave off fine degrees of material. In reality such things as machine rigidity can come into play, as a less rigid machine will be pushed off with a regular reduction cut, but a reduced depth final cut will not push the tool away so much, so can result in the workpiece going undersize for no readily apparent reason.

So we are back to practice to gain an understanding of what the nature of these cutting properties are on our given machines, in as many circumstances as we can, do enable us to develop a sense of what might need to be done in the way of machine setup and operation to achieve the desired results.

Yeah the trapped rule is a good one.

For others, what Tenn is doing here is to lightly trap a rule between the cutting tool and the workpiece, if an external tool is over height the top of the rule will lean towards the back of the machine, for boring tools use the back side of a test piece and the rule will lean towards you at the top if over height. The opposite in each case if under centre height.

I also agree what's said elsewhere, once a basic setting is in place, face off to judge an OD tool's setting, what we call a 'tit' will result if the tool is over or under, and it's a way of ensuring correct tool height when actually cutting.

For your interest, set a dial gauge on to a tool and zero it, take a cut and see what happens during the cut.