Education policy, like so many other things, is normally composed of a few overarching principles, onto which the details are fitted. And perhaps surprisingly, these few fundamental questions can be answered independently of each other. Pretty much any combination of these can be made into an educational system.
One question is what/when should the children learn. Should everyone have to know how to derive and integrate, or is it sufficient if future engineers learn it at university? Should there be more national "culture" such as literature, music, national history, or rather should there be more science, world history, perhaps a bit of economics?
A completely separate question is what skills do we want them to have? Foreign languages, "digital literacy", and so on. This is a separate question, because skills have a how added to them. Skills can only be acquired through practice—and what are these skills to be honed on?
The current answer is to treat skill-centered subjects the same as knowledge-based subjects, in isolation from all other subjects. This means that skill-based subjects have to come up with their own problems to practise on, and the palpable artificialness of these problems is very demotivating to the students.
My answer would be to "cross" skill-based and knowledge-based subjects with each other. For example, science classes work perfectly well with informatics. Several classes are culture-independent, and thus can provide problems to talk about in foreign languages. Naturally, the more thoroughly the knowledge-based and skill-centered subjects are meshed together, the better.
Another separate question is what values and behaviours do we want to transmit? The old answer was clear: order and obedience, because if children are closely regulated from six years of age, the military will have an easy time with them. To a degree, this value is still present in education: the strictly rectangular seating, cooperation being forbidden (don't talk in the back!), the regular length of lessons (that not even the teacher can control), etc. It's time to throw this out completely, not just piecemeal.
And I suppose we should bring in the transmission of knowledge itself, not just as the goal of the education system, but as a value it transmits. We should build the practice of mentoring into education. And I don't just mean teacher-to-student, but—for instance, above year 6—between students of different age. In the last, say, 2 years of high school, from university students. In universities, from older students at first, and then from professionals working on real problems. I.e. professors who work in industry.
It would even benefit companies, if university professors wouldn't be full-time professors, but primarily would work in industry and only teach at the university part-time. After all, they would bring the smartest students to internships, and if they do well, employees. This would largely bypass the difficulties of trying to recruit from the market, because the most talented people don't show up there, while the unqualified and thus often-unemployed make more noise. This would also bypass the risk of giving the job to someone unfit, and then having to fire them.
More importantly however, the above would greatly benefit students (who would pass on the practical knowledge to younger students via mentoring) and society. The latter is due to ingraining the culture of sharing knowledge.
It is widely known that the industries with the highest value added always are what counts as high-tech. We can also observe that world-class innovations, that make it to industrial practice, usually come from large countries.
It is worth asking the question, why small states don't have their own high-tech industries? Of course, there are the economies of scale of production. However, this is not the real cause; if they can license something developed abroad, they can supply themselves just fine, there are many examples of this. The real cause is that having any kind of high-tech industry takes a large amount of knowledge. You need a group of engineers, who need to develop several versions before it doesn't merely work, but works well. Then a significant quantity has to be made, to find out manufacturing and operating issues. The additional experience gathered can then be incorporated into the next version, which again needs to be manufactured in sufficient quantity to learn from it.
However, knowledge has to be built up not just on the development side, but also the operation side. When something breaks, it cannot be reparied efficiently if the fixit guy has to reverse engineer it on the spot. To put it differently, there need to be enough repairmen so one can be reached when necessary. This is only possible if the thing is used widely enough, and so its ways of breaking down come up frequently enough for there to be repairmen who are familiar with it.
And it is this amount of knowledge, this group of skilled people is what the economy of a small country cannot afford. And it is why manufacturing does not pose a problem, once the design work has been done abroad. And why maintenance so often does pose a problem.
The previous train of thought can be applied to companies, too. If there are N small companies, that each jealously guard their own little intellectual property, that will not result in world-class development. However, if in one area—both in the geographic and the specialization sense—a culture develops, whereby several companies with similar profiles cluster, so engineers move back and forth between them, and transfer their knowledge among each other, then they can build up a large body of experience, and a name like Silicon Valley emerges.
Sadly, this culture is its own enemy. As soon as it becomes successful, the foundation of its success—on the level of the world economy—will be that they know something others don't. Even individual companies are more successful than the others when and while their engineers come up with something the others don't yet know. As soon as the others hear of it, they can again compete on equal footage, and this drives down profits for the first company.
Thus as soon as the culture of sharing knowledge becomes successful, it is in the interest of individual companies not to share their newest results with the rest of the community. If they can act on that, the successful culture is replaced by secrecy, and the drive of success shuts down.
If this happens while there are multiple small companies, then they all slowly sink back to being average. That is because none of them can finance in-house the mass of knowledge required.
If this happens when one company has taken over much of the market, theoretically they have the ability to pay for all the costs of development in-house. In practice, though, if they are unquestionably the largest, and the others cannot keep up with them, they can slow down the pace of development, and can cut corners on the body of knowledge, to increase profit.
This is the widely known effect monopolies have on development, with the added complication that the small companies surrounding the hegemon are unable to pay for the price of innovation, thus this effect can be in play without an actual monopoly.