Reality check: Is this a plausible way to breed raccoons?

You're a lone immortal who's decided to breed raccoons to a human-level intelligence (over the course of hundreds of thousands or millions of years). To do so, you need to both feed a significant population of raccoons, and select for intelligence & sociability.

You're hoping to maintain a population of about a thousand raccoons, though more is certainly better. North America was likely settled by only about 70 people; human populations likely went through a number of bottlenecks, and genetic evidence suggests that human population at one point may have been as low as a thousand individuals, allowing for rapid differentiation. 1000 raccoons is therefore probably a reasonable target.

A large raccoon will weigh about 8kg. Assuming a raccoon needs about the same amount of food as a cat, they will need 13 kCal per kg per day, and so you need to produce 104,000 kCal per day to feed all the raccoons. Crickets are a good way to get this: there are 1.21 kCal per gram of cricket. One 20x20x20cm breeder box can produce 125 grams of crickets in 36 days. This means 4.2 kCal of cricket per box per day. This means you need to have about 25,000 breeder boxes, or 1000 square meters. To feed the crickets in turn, you need on the order of 150kg of biomass per day. Luckily, crickets aren't picky, and the setting -- Northern California -- is filled with fertile ground. This seems like a plausible amount to collect as an individual, and some of the raccoons can eventually be trained to collect biomass and dump it in the cricket pit. You may do a little small-scale agriculture as well to supplement their diet -- apples could be used to reward raccoons for especially good behaviour -- but the bulk of their calorie intake will consist of crickets.

The raccoons will be kept in pens at first. Since raccoons are solitary by default, the pens will be individual initially, but as you breed your raccoons to be more sociable they will start living communally. Young will stay with their mothers for the first year, then be transferred into their own pens.

Food, in the form of crickets, will be placed inside a variety of puzzles. These puzzles will slowly increase in difficulty through the generations, and a fair bit of your time will be spent thinking up and building these puzzles. A simple puzzle might be needing to pull open a drawer. Generations down the line, these puzzles might be math problems. Given the problem 2+2, the raccoon must place a rock in the fourth slot. Other puzzles might require teamwork. You will give out some food personally as well: traits like friendliness will be rewarded.

Food will also be the proxy by which the raccoons are judged. Any raccoon that's looking too skinny will be removed from the breeding population by releasing it into the wild. While most of these exiled raccoons will die, those that survive will disperse the genes you've been selecting for into the wild.

Genetic diversity will be maintained by regularly catching raccoons from the wild and inserting them into the population you're raising: a dozen or so each year. You will also occasionally release a dozen of your smarter raccoons each year. If all of your raccoons are killed, hopefully you won't have to restart from nothing.

Is this a plausible way to breed raccoons? If not, what are the flaws?

---

Further details:

• I'm not interested in natural disasters, eg earthquakes, destroying the pens (these natural disasters form a plot point)
• I only care about the relatively near term. As my raccoons grow in intelligence, the strategies will shift, so for the purposes of this question, these are pretty normal raccoons. In other words, you don't have to worry about a raccoon uprising.
• I'm most interested in structural problems that mean I'm not selecting for smart & social raccoons, and things that might kill off all my raccoons unrecoverably

Feel free to challenge this, but expected technology level is likely to be more or less medieval, since there won't be the infrastructure for any precision machining:

• Will be straightforward to do metalworking with copper, since there's plenty of raw material in the form of wires and it's been worked since ancient times
• Should be possible to work with irons and steels as a solitary blacksmith
• Can make concrete, but rebar will be prohibitively time consuming
• You know modern chemistry, biology, and other sciences (reading & preserving this knowledge is one of the first things you do), but long-term you will only have the equipment you can produce yourself

---

It's been nearly a year since all the humans on the planet mysteriously disappeared, leaving you alone in a decaying landscape. You are the only human left on earth, and you are immortal. There are a handful of tantalising clues scattered about labs in the San Francisco Bay Area. What few you can decipher point toward a massive puzzle too big to tackle on your own. You need other minds working on the problem, and the only way to get them is to make them: with biology. Knowing full well that it might take millions of years, you set out to selectively breed a new species to intelligence.

You don't need to eat or drink, although you do feel hungry and thirsty. You do need to sleep for at least a few hours each night. Psychologically you're fine, and even if you're a bit lonely, you're not going to go insane. Your memories do fade, but not past the point of a memory from five years ago or so. You can be wounded, but your wounds heal almost instantly. If a limb is severed, it dissolves and re-forms. No diseases can affect you, and physically you're a 25 year old in peak health. You're smart and well educated, but you aren't a true genius. Even if you want to, you cannot die.

Followup to Best species to breed to intelligence