While in fluorescent lamps, the electrodes are either preheated before ignition (By a glow starter or electronic starter in preheat/switchstart magnetic ballast and without it with the israeli rapidstart magnetic ballasts, PTC and programmed start electronic ballasts), or heated during the ignition (With the american HPF rapidstarts and UK trigger-starts), in LPS lamps the ignition is usually done in a same manner as the HID lamps by using a semi-parallel or a superimposed ignitors, without any electrodes heating.
This despite that both the LPSs and the fluorescents are low pressure discharge lamps, with large tubes.
Also the electrodes of the LPS, although have a different shape of a beehive, are still filaments with two wires, similar to the fluorescents electrodes, meaning that they can be heated.
Because this type of ignition of the LPS lamps, considered a cold cathode starting of a non cold cathode lamp (LPS lamps still have filament electrodes, and so they should also be considered thermionic emission discharge lamps), each time the lamps ignited, a massive amount of emitter is sputtered, and as so, frequently switching of LPS lamps, shorts their life considerably, similar to fluorescents operated by a low cost instant start electronic ballasts, and probably even more then them.
Why this is the case? Why LPS lamps ignited in the same manner as the HID lamps and not like the fluorescent lamps?

I suppose the shape of the SOX lamp makes it impossible really, you'd need some sort of special 4-pin cap and external starter for preheating. The longer higher wattage lamps are normally continuously pulsed by the ignitor during warm up or they just extinguish so they wouldn't work on a standard switchstart circuits anyway.

The linear SLI low pressure sodium lamps except the 200W high output, were used on switchstart circuits though.

I could be wrong, but I think it largely boils down to manufacturing costs and intended use.

The "start cycle" limitation due to the lamps being started cold cathode isn't likely to be a massive issue in applications such as streetlighting where the lamps are only likely to be switched once every 24 hours. 

The simplification in control gear and lamp design is probably seen as a greater benefit than the disadvantage that is the potential lifetime reduction.

I think Zelandeth is right - when more frequent starting then once per 10 hours is not necessary, cold starting does not cause any significant life degradation, so the cheapest setup (in the early 50's or earlier point of view, when standards for these lamps were made; so integrate glowbottle starter into the socket was technology still about 30years out of reach).
The cold-starting of SOX mean great simplification to lamp and it's holder, however the ballast become more troublesome (has to have high OCV or pulsing ignitor; first add dissipation, second total complexity), but the complication with 4-pin socket was likely considered too much and too expensive.

Early low voltage LPS lamps did use cathode preheating, and worked in a manner similar to North American rapid start lamps, where the electrodes were fed with a constant heater current.  When a high voltage style was adopted, it was found that the discharge kept the electrodes warm enough, and the additional windings and contacts for cathode heater current wasn't necessary.  This is also at a time before simple and cheap glowbottle starters for fluorescents existed (and about 6 years before fluorescents themselves were commercially available).

And with a light that takes 8-12 minutes to warm to full brightness, frequent switching isn't really a big issue, and like Zelandeth said, their intended use was dusk to dawn, with one start per night.

Another reason, street lights on long outreach arms tended to shake in the wind, so lamp makers have to make a more robust cathode than in flourescents.
The most common idea is to `wrap` the coiled tungsten wire around an armature wire.
On start up, on a flourescent type pre-heat circuit, the pre-heat current would not go through the tungsten wire, but rather through the `heavier` armature wire causing a dead short.