Is dietary carbohydrate even necessary at all? In the recent past, this question wouldn't even be asked. Knowing that the body can manufacture its own glucose via gluconeogenesis, dietary carbohydrate has long been considered non-essential. Recently, however, The Jaminets, in The Perfect Health Diet (PHD), and on their website, have advocated getting glucose needs met through direct consumption, and express the view that glucose deficiency is imminent in those who consume fewer than 50g of carbohydrate. The PHD recommends 400-600 calories of carbohydrate per day, or 100-150g.
In contradistinction to this recommendation, Phinney and Volek, in The Art and Science of Low Carbohydrate Living (ASLCL), describe a functional gap in human physiology, at precisely this range of carbohydrate intake. The argument goes that if enough carbohydrate is consumed to suppress ketogenesis -- more than 50g -- then at this range you have barely enough glucose to feed your brain and no alternative source of fuel. The only way to support any extra glucose needs under this condition, such as may be needed by extra activity, is to consciously notice how much glucose you need and supply it at frequent intervals. On the other hand, if one consumes little enough carbohydrate for long enough that ketoadaptation occurs, then the brain's needs can mostly be met by ketones, and moreover, most of the muscle fueling will be provided by fat. Glucose needs drop tremendously in response to not eating any.
At first glance, it may seem intuitive that the best dietary intake of glucose is to predict and match the body's glucose needs. This would "save" the body from having to manufacture any. There are problems with this idea, however. First of all, just because glucose needs to be manufactured, if not provided, and this is obviously not without cost, there is no evidence that the process is difficult or "stressful". (See http://paleohacks.com/questions/25449/why-do-people-consider-ketosis-stressful-to-the-body.) And ingesting glucose for use is not "free" either; to the contrary, the knock on effects of ingesting glucose -- the effects of raised blood glucose, as well as the effects of the raised insulin that is rushed in to cope with the raised glucose, and the extra oxidative stress involved in carbohydrate metabolism -- may be much more "stressful", than the gluconeogenetic process.
Moreover, guessing at precise carbohydrate need seems like an error-prone pursuit. The risks of underestimation have already been mentioned, but overestimating may also be costly. Carbohydrate is a "preferred" fuel, which is an amusing and misleading anthropomorphic way of saying that the body needs to get rid of it right away: when it's available it will always be used first, because to do otherwise would be harmful. Using it may also be harmful, in comparison with using fat or ketones, at least insofar as it creates more oxidative stress and inflammation.
After considering the possible pitfalls of trying to estimate and provide for the precise glucose needs of the body, a questionable evolutionary strategy, it seems more plausible, at least to me, that the best method of getting what glucose you need, is the hands-off approach of letting your liver (and kidneys) make it on an as needed basis, responding directly to internal fuel needs. The remaining question that needs to be answered is whether this is adequate. Can this approach really lead to a so-called "glucose deficiency."
I want to pause here and stress that there is a big difference in saying that it is implausible that the body cannot meet its glucose needs without dietary glucose, and saying that the glucose deficiency symptoms that Jaminet has outlined don't exist. It would be foolish (not to mention disrepectful) to ignore the real experience that some people, including Jaminet himself have reported, just because it doesn't fit a theoretical framework. To quote Namby Pamby (a comment to this answer: http://paleohacks.com/questions/68656/dr-ron-rosedales-recent-posts-about-safe-starches-what-do-you-think/68748#68748), some people get the following symptoms on ketogenic levels of carbohydrate, that disappear with sufficient added starch: "constipation due to mucosal dryness; fungal toes and scaly feet that resembled an alligator's; dry eyes (although for me this may have other causes); fainting spells when getting up from a chair; possible hypothyroidism, as I display subclinical symptoms; and low LIBIDO!" So my question is, what is really going on here? I want to draw an analogy that might be appreciated in the paleosphere. It is indisputable that low carb diets can reverse metabolic syndrome in a large proportion of affected people. Does this mean that carbohydrate consumption is the cause of metabolic syndrome? No, it doesn't imply that. Similarly, it may be the case that Jaminet's Syndrome can be reversed by eating 150 grams of carbohydrate a day. Does this mean that lack of dietary glucose causes Jaminet's Syndrome? No, it doesn't imply that either.
I haven't read the PHD (I'd like to!), but based on this blog post, the Jaminets' belief that gluconeogenesis cannot provide all the glucose we need comes from adding to the brain's requirements at least the following: glycosolated proteins, glycoproteins, immune function (to create ROS's), feeding infections, and athletic activity.
I'd like to focus on the point about feeding infections, because it is unique in being not necessarily a normal body function, and therefore it is plausible that the body didn't evolve with this functional requirement. To summarize an idea that I think also comes to me from Jaminet, the character of infectious disease has, in recent decades(?) changed from the type that acutely affects its host, often killing him, to the type that lives like a parasite, crippling the function of the individual to a greater or lesser degree. So it may well be the case that a significant number of modern folk are carrying with them an infection that needs feeding. And it seems plausible to me that such beasts could steal glucose from the system that cannot be replaced quickly enough to avoid problems in the host who is not supplementing with carbohydrates. Moreover, there is a rumour I've been hearing that I haven't verified, that fungal infections are particularly insidious, because they are not weakened by glucose deprivation, being able to survive on ketones as well. If this is true, and a fungal infection is strongly present in an individual, then it could explain why this person's health would worsen, rather than improve, on a ketogenic diet, and specifically that they would have fungal, scaly skin as a telltale symptom.
So what I am proposing is that the preponderance of people who are not faring well on a ketogenic diet, but who thrive on a middling level of carbohydrates, which otherwise, it may be argued, should be the worst possible range, is that there is an infectious epidemic that is impairing many bodies' natural abilities to provide enough glucose. If this is correct, then being constrained to eat starch is a condition similar to the condition of being constrained to eat a ketogenic diet for the "metabolically damaged", i.e. those with metabolic syndrome. A possible test of this hypothesis is that if the infection could be identified and cured, the person would be expected to then tolerate a ketogenic diet without experiencing glucose deficiency.