I agree. SMD is the first choice for small modules like the ESP8266 WiFi module or others. For hand soldering and debugging THT is much better and easier to handle. My biggest horror is to have a expensive large system completly build with SMD parts that is broken by electric discharge for example (my ZX-UNO VGA 2M has a short circuit situation, all SMD - how to find the defective part and repair ?. SMD parts are very sensetive against tin oxidation, they have to be stored encapsulated to prevent this. THT parts are much easy to handle regarding this. Generally I'm too old to work with these small parts... I'm anyway allways searching for my glasses

I have a Metcal soldering station; it is a regular solder station with interchangeable tips. For SMT I use STTC026 tip which is 0.4mm diameter.

Inspection microscope has long working distance so you are not too close to the fume which is vaporized rosin. A small fan help, but for occasional work I don't see the point. Lead/tin solder is in continue use in USA. Because of potential tin whisker formation over long period of time, high reliability electronic still require tin/lead solder and are generally hand soldered because of small production run. The assembly persons are almost always old ladies with 20-30 years of experience handling lead/tin solder everyday. I know that because I worked with high reliability assembly for many years. I also know if there are any health issues with handling lead/tin, these old ladies will make a fuss and they haven't.
Bill

the dandruff that claims to be capacitors and resistors is just beyond my eyesight even with a magnifier.

• THT can indeed be made very dense, particularly in analog circuits where you use a lot of resistors and diodes that can be staggered in up to five rows underneath the DIPs (so they take no board space at all!). I used to do this for designs at work all the time. We were very late in going to SMT, because we had our own assembly and soldering machines so we didn't have to farm it out, and the set-up costs for automated SMT assembly were still quite high for the small quantities we produced. The difficulty in troubleshooting and repairing a board with parts under parts actually made the production people so careful that we had almost no errors. I got across-the-board densities up to 45 parts per square inch this way. I realize this is not practical for hobbyists though.
  
  
• You can indeed put parts on both sides (more likely in digital circuits, not analog which has a lot of discrete components) if you stagger the rows of pins. It works best with open-frame sockets (or strips) so you can solder a socket in place and still have access to solder an IC or socket that comes in from the other side of the board, before covering those pins with the IC that goes in the socket. Another way is to solder the pins from the top, although you might not be able to put the ICs quite as close together and still get enough access. The soldering iron only needs to touch the top of the pin, near the body of the IC; but you still need to get the solder down near the board.
  
  
• A minor problem with getting density with parts on both sides with thru-hole is that every hole takes space on every layer of the board, not only complicating the possibility of parts on both sides, but also making routing of signals more difficult on internal layers. (Offsetting that slightly is the problem of putting vias in pads in SMT.) So another possibility is to either J-lead or I-butt thru-hole parts, and solder them like SMT, onto pads on the board, with no holes. J-leading is best done with a special tool (which I don't have). I-butting is where you cut the lead in the narrow part, a very short distance below the wide part, and then have the solder fillet fill in the overhang space.
  
  
• For work, I prefer designing with SMT, for reasons of density and AC behavior. I can solder SMT by hand, but it really is a problem when we're getting a new product going and we have to troubleshoot problems and show the assembly house or board manufacturer what they did wrong. We do have SMT rework equipment, but it's just so much easier in thru-hole. SMT and automated assembly works great once the bugs are ironed out and the product is launched. The problems we have are on very few boards of a batch, meaning it's not the design; so sometimes I'm tempted to just tell the manager it's not worth my time to try to troubleshoot the thing, and he should just accept the loss of a small percentage of boards that don't pass the testing. The recent problems I can think of were that they cracked an 0402 resistor (the crack being visible only with a microscope), knocked another one off apparently because the solder was already cracked (a problem which does not happen with leaded solder!), and had a crack in a trace near a pad, and had one connection open at the side of a tiny module because they did not use enough solder and flux.
  
  
• For things like connectors where there's stress on the pins when something is plugged in or unplugged, I greatly prefer the strength you get with thru-hole and plated-thru holes, where it's like the pin is riveted in. I've seen too many times where SMT connectors peeled the foils away from the board when stressed.