A potentiometer is a variable resistor. Pots are the primary choice of transducer for converting mechanical position into an electrical signal that can be used by your sim. When you pull back on the yoke or push the throttle forward, you're changing the setting (and the resistance) of a pot.
A pot generally has three wires or terminals. Two are simply the connections to the ends of the resistive element. The remaining terminal connects to a moveable contact called the wiper. The wiper slides along the surface of the resistive element as the pot's shaft is moved. As the wiper is moved closer to one end of the resistive element, the resistance between the wiper terminal and that end terminal decreases.
Some ancient history: Slidewires
One of very oldest versions of the potentiometer is the slide wire; basically just a length of resistance wire (like nichrome) with a sliding contact that can be positioned anywhere along the wire's length. Slide wires are simple, quite robust, but overly large. The resistance wire can be wound around an insulating tube (often ceramic) to make the unit more compact. Slide wires are often seen in freshman introductory electronics labs and old Flash Gordon movies.
A very close cousin of the slide wire is the wire wound potentiometer. (High power wire wound pots are referred to as "rheostats".) Because the characteristics of resistance wire can be very closely controlled, wire wound pots can be made very precisely and very linear. In fact, precision grade pots are commonly wire wound. Manufacturing wire wound units is somewhat more involved than manufacturing other types so it's not too surprising that the cost is generally somewhat higher. Some models are produced with substantially upgraded lifetime and reliability that of course results in a substantially upgraded price along with the substantially upgraded name "resistive position sensor". Generally, you will not find wire wound potentiometers made in very high resistance values.
Some slightly more recent history: Carbon Comp Pots
Although not quite so old as the slide wire, carbon composition pots have their roots firmly embedded in the mists of time as well. While you won't notice many in old Flash Gordon movies, it's interesting to note that the audio system in the Ark used quite a number. Carbon composition resistive elements are a combination of finely ground carbon granules, clay-like insulating filler, and a binder to hold everything together. This mixture in varying proportions produces widely differing resistance values. Values as low as a few hundred ohms to as high as several megohms are possible. Unfortunately, some composite resistive mixtures can be quite susceptible to moisture with the result that values will shift when pots are exposed to changing humidity. So, it should come as no surprise that after a few weeks aboard, Noah's wife complained that their audio system sounded worse than the Ark smelled.
Wear -> Dust -> Noise
As the carbon comp pot wears, granules of the resistive composite collect under the wiper causing it to have erratic contact. This is the noisy pot syndrome, and in your sim will show up as jumpiness in the axis the pot controls. This jumpiness can be partially controlled by electrically connecting the wiper to the unused terminal of the pot. This won't eliminate the jumpiness, but it will limit it somewhat. (Please note that this is not the only reason for jumpiness. Electrical noise coupled into unshielded wiring to the pot is another possible cause.)
In an effort to improve the performance of pots, materials people developed cermet and conductive plastic elements. Cermet is a mixture of ceramic and metal powder that is fired together to product a resistive material that is more stable than carbon composition. Conductive plastic offers another, more stable alternative to carbon comp.
Tapers: linear & logarithmic
And now for a few more bits of potentiometric lore… First, they come in differing "tapers". A linear taper means the resistance measured from the wiper to one of the end terminals will vary linearly as the pot's shaft is moved. You will also see pots that have a logarithmic taper or a reverse logarithmic taper. This means there is a highly non-linear change of resistance as the shaft is moved. These tapers were developed for use in audio equipment to match the non-linear human perception of loudness. For this reason you may see these tapers referred to as "audio" tapers.
Potentiometers are available in a variety of physical forms. You're perhaps most familiar with the single turn rotary shaft type commonly used for volume controls. Actually, they aren't precisely "single turn" types. If one were to be obnoxiously accurate, they would be "three-quarter turn" potentiometers because internal stops limit the rotation of the shaft to about 270 degrees. The resistive element forms an incomplete loop with a portion of the ends of the loop taken up with the non-resistive contacts that tie the element to the terminals. They are variations on the rotary pot that provide 3, 5, 10, or even 20 turns of the shaft which are particularly useful in providing more precision in setting the resistance value. It's not unusual to find that the prices are 3, 5, 10 or even 20 times the price of their "single" turn cousins.
A caution about rotational range: A "linear" potentiometer with 270 degrees of shaft rotation is not truly linear over all 270 degrees. When turned to the full left or full right position, the potentiometer's wiper is parked over the area where the resistive element makes a right angle turn and is connected to the metal tabs you solder to. The change in resistance as you adjust the potentiometer near the ends is decidedly non-linear. If you are depending on good linearity over the operational range of the potentiometer, avoid operating too close to the stops. Ten to 20 degrees on each end can be taken up with this non-linearity in the connection area. So a 270 degree linear potentiometer may really only offer 240 degrees of linear movement.
There are also slide potentiometers that are controlled with a straight-line movement rather than the twisting movement of a rotary pot.
Resistance & power
Unless you look carefully, you're likely to see only two numbers describing the pot (excluding the price, which would make three). The first number is the resistance. Typically, a value of 100k ohms is used in joysticks. The second number is the power rating. Power handling capability is essentially of no consequence for joysticks. There is no value in using a 2 watt unit rather than a tenth watt one other than perhaps a small gain in physical robustness.
And lifetime & tolerance & tempco
There are actually three more numbers that are worth being aware of, and one of them is actually worth looking for. And that one is the lifetime. It's generally expressed in the number of full rotations. A general duty pot used for something like an audio volume control might carry a rating of 30,000 to 50,000 rotations. I wouldn't recommend a pot with a smaller rating. If you have problems wearing out pots, you might want to invest in pots with 1,000,000 rotation lifetime specs. This will probably be a cermet or conductive plastic type.
The second of the three numbers is the resistance value tolerance. This might be anywhere from +- 10% to +-30% for the less expensive models. (There are some very precise units available, but they are quite expensive and not warranted in these applications.) If you are not building controls that have to track each other, don't worry about tolerance. Even if the controls do need to track (for instance, quad throttles) it still may not matter if your joystick setup and calibration software can handle the difference. The third number is more important.
This third and final number is the temperature coefficient of the resistance element. It's a measure of how stable the resistance will be as the temperature changes. Smaller is better. It's generally expressed in parts per million change in resistance per degree change in temperature. The interesting thing is that if you can find a tempco rating, it's probably fine. The loosest rating I've seen in catalogs is +-250PPM/°C. For a ten degree C change that would only be a quarter percent change in resistance, not a big deal. If there's no rating at all, who knows what the pot's going to do? Poor tempco is an issue in applications that have to track, like the throttle example.
Practically speaking, what to buy?
So, all right already, what's the bottom line? What should you buy?
Dunno… Guess it depends what you're gonna use it for.
If you're planning on using it for your flight controls, here are some thoughts. First, you want a linear taper. The typical value found in a joystick is 100k ohms. (Caveat: if you're replacing a pot in a USB unit, it may not be 100k. Check the old pot to be sure.) Next, unless you have some funky gearing arrangement, you will likely want to stick with a single turn pot. Possibly you're thinking that it might be easier to mechanically connect your homemade throttle linkage to a slide pot rather than to a rotary pot. A consideration is that it's easier for the manufacturer to seal a rotary pot against dust and dirt than to seal a slide pot. Dust and dirt adversely affect a pot's lifetime.
Wire wound pots offer superior precision and linearity, as you will recall from my earlier scholarly writings, so probably you're thinking that's what you should get. Actually, I neglected to point out a small, tiny, almost insignificant shortcoming of wire wound units. They're not continuous. As the wiper is moved, it makes connections to adjacent turns of resistance wire, rather than sliding over the whole length of the wire, so the output actually takes a lot of tiny steps in resistance. While its output is considered linear, it is not smooth. Depending on the particular model of pot, this may or may not make a difference in your application. But, as it turns out, you really don't need the wire wound pot's precision or linearity, so don't spend your money on it. (Remember, we're talking about non-geared, flight control usage. Other applications may greatly benefit from wire wounds.)
The bottom line, or at least A bottom line
A good quality carbon composition (or carbon film) potentiometer will do fine. Cermet and conductive plastic would work quite well too but cost more. All of these choices have continuous resistive elements that offer continuous resolution, which is what you want.