I agree with some of what this poster has said. There are a lot of people graduating each year, more than I'd like to see, and there are a limited number of openings. There have already been several good posts in this thread about advanced degrees and flexibility in relocating.
I do take issue with a couple of points in the original post.
1. While I can't speak about stress in the private sector, I can say with some authority that the stress level at the NWS field level varies with location and time of day. At our office, it is low on "Big Bubble" days...and extreme on active weather days. It can be especially stressful in the short time we have between data availability and forecast/discussion/statement deadline time. The baseline stress level also goes up everytime Headquarters has a great new idea to add to the (already saturated) workload.
2. The Graphical Forecast Editior in the NWS is not a panacea. We aim to add value to the raw data, and that means doing more than simply "plug and chug". We go through the model data as thoroughly as we have time to do (see #1), and frequently find the best solution is a blend of two or more models or versions of models (00z vs 12z). Some of the time even this isn't enough, and we need to hand-massage the grid fields. Considering that there are roughly 40 different meteorological fields, many of which are hourly (24x7 = 168 hourly grids per field), that is a lot of data to cover. We also need to QC all of these grids to ensure we avoid "Mostly sunny with a 50 percent chance of rain." QC is not done automatically by the system...when we change even one grid, we need to check all the fields that grid may affect. So while the NWS domains may be "small", it is not accurate to imply they are easy. And the forecasts, while not hand-written, still require a heck of a lot of work.
3. An example on grid editing and workload: Our office has been working on grid-generated aviation forecasts (the last and most difficult forecast type to grid-ize). This will allow us to produce forecasts for many more sites than we already do...and we already forecast for 10 sites (second only to LOX in the lower 48 states). To be sure, we do spend less time typing aviation forecasts now. But when you add in all the grid editing and QC, it still takes the same amount of time to produce a forecast package as before. So automation has shifted the workload, but it hasn't reduced it.
3. Unless there is a very quiet weather day and week, we do not even consider having one desk handling the entire forecast. Too much detail is needed.
4. NWS staffing has 8 to 10 forecasters per office. On a technical level, the administrative team (MIC, SOO, WCM, Service Hydrologist) may be able to fill in...but some offices have local policies that discourage this. Some Service Hydrologists are degreed Hydrologists, not Meteorologists, and so do not staff a weather forecast desk. The Public Service Desk may have Interns, HMTs, or a mix of the two; while the Interns may be able to fill in from time to time, that is not what they are there to do. So quoting "122 offices, each with about 20 people" makes us sound a lot more flush with bodies that we really are.
The NWS hasn't been "easy" to get into since at least the mid '90s. So I urge everyone contemplating a career in Meteorology to think long and hard about it. And if you do feel a calling for this field, have a Plan B. Given the current economy, a Plan C and Plan D probably wouldn't hurt as well.