I couldn't agree more with the basic gist of the linked post; if I am going to make a random chart at all, it has to be useful in a thousand situations and not stale after six rolls. Of course, Tao goes beyond simple complexity and demands of himself (and others, I guess, if I were to acknowledge the sub-plot of many of Tao's recent posts) charts and game mechanics that model the consistencies and vagaries of real life within an acceptable statistical deviance.
I usually just stick with making charts with a gajillion results. Ferrinstance, when I wanted to randomly generate magical fungi for my Mutant Future campaign's brief stint on Celestia, I created a "chart" with some number of thousand total possibilities,and I actually do not even possess the statistical acumen to give you a correct total. The version that I posted on the interwebs is lacking the addiction and overdose tables, but even just the primary effect (d30 + subcharts + some random duration rolls) + secondary effect (d20 + subcharts + some random duration rolls) tables generate thousands of possibilities.
That is the kind of chart I am going to create, if I create any at all, because in general I just rely on my ability to perform well under pressure and come up with something awesome, and my ability to sit back at the correct moments to allow my players to come up with something awesome. The time that I spend in prep is normally spent creating new creatures and cultural practices and items and sketching and most dearly refining the physical rules that govern my RPG universe (i.e. expanding on magic and its relationship to energy, time and intent). It is spent jotting down quick reminder lists of the various NPCs and factions that are going to be most active during the period of time the session is likely to entail. When multiple days pass in session, I rely on my own ability to just create weather conditions and terrain descriptions that are not just accurate for the latitude and time of year, but evocative as well. I do a fair amount of research on conditions in Earth analogs of the kind of places I stage my adventures, and I just wing it from there. It seems to work well; weather is not an afterthought, it is a tangible part of the game, but I don't think any of my players care HOW I generate it.
I fly without a safety net a lot of the time; not all of the time; not even most of the time; and my poker face is good. I act the same one way or the other if I am improvising from the seat of my pants based on the setting to date and the actions of the players, or if I am working from fastidiously prepared notes.
Don't get me wrong; I DO have fastidiously prepared notes. I prepare reams of material for every game I have ever DMed... scribbled pages upon pages of notes, pen and ink, illustrations, diagrams, sidebars and more. I just accept that a good portion of that work is just a mental exercise and that the number of hours spent conducting freestyle mental gymnastics during a session has no correlation to the number of hours spent preparing for a session.
There is rarely a situation in one of my ongoing games that I feel the need to create a chart for, because I have taught myself the lesson over and over again that I can be a successful DM just by staying out of my own way; providing interesting tidbits and reacting to player actions. I have honestly never felt the need for "random" encounters and weather, and even when I have made such charts I usually have ended up using them as inspiration in the moment. A post like Tao's makes me sort of wish that I had an alternate reality version of myself to make awesome detailed charts for every region in my 4e game, while the real-life me kept on happily creating undead mechanical constructs, strange tribal customs and new magic. But then I remind myself that Tao's campaign is set in a world that exists, Earth, and a history that not only mostly exists (late middle ages with magic, if I can sloppily summarize), but is familiar to the typical gamer, while my games tend to be set in very alien climates and social conditions. I fall closer to the Tekumel side of the divide than the Pendragon. I align myself nearer to the Sci Fi / Arnesonian / Temple of the Frog school than the Fantasy / Gygax / Keep on the Borderlands camp.
There is something about your standard, bog simple, pseudo-European generic default D&D setting that does still stir my blood, I must admit. I am really enjoying what I am working on right now precisely because of that. I am making random encounter charts for my entry into the Iron Chef Adventure Challenge issued by Mike Monaco (Swords & Dorkery). There are going to be several layers of complexity layered on top of your typical THIS kind of terrain, THAT chart routine. The basic format is going to be a d20 roll for each terrain type that a party could travel through in the hex-crawl (currently: Farmland (grain - travelling by road); Farmland (grain - cross country); Farmland (root crop - travelling by road); Farmland (root crop - cross country); High Moor (via road & traveler's shelter); High Moor (cross country); Millbrook Town; Hot Springs and Abandoned Estates surrounding Millbrook; Mixed New Growth Forest and Farmland; New Growth Forest; Old Growth Forest; Mountain Foothills; High Mountains; Glacial Ice Field; Temple Marsh & Hotsprings; the Moon; the Sun; Astral Plane) with a result of 20 calling for a roll on a special chart.
I am going to call for a roll once per hex entered OR once per day if no travelling is conducted (with a secondary roll for time of encounter) .
Results 1-15 on the d20 terrain type encounter chart typically call for a roll on a sub-chart (a common feature of my chart design; when creating a version of Mutant Future for "Fantasy" gaming, I used the [d100 roll gives a result that leads to a sub-chart] mechanic to create WAY more variation in random character generation than the original game's mechanic); for instance, these results might be "Wildlife - roll 3d6 on the following chart" or "Civilized Folk - roll 2d12 on the following chart" or "Uninhabited Structure - roll 1d20". Results 16-19 are more unusual / notable encounters for that terrain type. A result of 20 = a roll on the special chart for that hex. Most hexes within a terrain type share the same special results chart, but many hexes have unique charts.
The special charts are d30 charts and the linear distribution means that once the d30 comes out, some crazy results can come up. Results 25-30 are typically major encounters that occur on a spread of hexes, with Demonic activity concentrated in the areas near the temple but still occurring on a 30 in hexes all the way out to Millbrook.
On top of that, the locations of some major NPC actors will be rolled randomly when the PCs enter the region and then their subsequent movements determined on a daily basis on another chart.
On top of that, there is a d100 roll each day for the 1% likelihood of a spontaneous portal to a (random) extra-planar destination opening in a randomly determined hex around the temple for a random amount of time, each portal with its own supplementary charts that reveal what crosses through to this plane, and subsequent movements.
I would love to add yet another layer of complexity, the ability for the last result to influence the next result, but I am not sure the players would pick up on it. It does give me something to aspire to, if for no one else but myself! There is no reason to be lazy with chart creation. The entire purpose of charts is to save you time during the game, so investing time in their creation will give an exponential reward as a DM. If you aren't going to go all out, then why not just ad lib it?