Well, having some sort of "market socialism" with employee-owned and -run businesses competing fairly in a regulated but still profit-driven market seems nice on its face, but as long as profit remains a direct goal, you can't escape the law of value. The USSR itself was heading towards the same fate even tho the companies weren't even employee-run.
A centrally planned (or collectively, democratically planned one, for that matter) doesn't seem to have a theoretical limit, only practical ones. There's a point where there's just plain too much information to tackle. A single person obviously can't process it, an army could process it but this introduces the crippling hurdle of communication between the people, which would completely deform the plan as it should have been. Obviously, this was the case with all existing centrally planned economies, and every single one of those suffered from the same problems we know so well: underproductivity, commodity scarcity and the infamous blat system. And since the cause of these issues is the communication between planners as opposed to planners' skills, you simply can't solve it by hiring more qualified and more planners. To fix that communication issue, either you a) devise far more complex economic modeling (assuming this is viable in the first place), which not only exponentially increase the amount of information to be handled, but also exponentially increase the chance of introduction of errors and the propagation thereof, or b) invent telepathy so all the planners can know exactly what the other is thinking or some shit.
A way to sidestep the communication problem would be to change the paradigm: abandon the market completely, and have the central plan detail the economy fully, down to the amounts of commodities being traded. The absurdly complex of planning an economy, which is arguably built around setting the right prices to commodities, would be simplified by simply dealings in amounts of commodities without any value or price abstrations, making calculations infinitely easier. This would entail the elimination of currency except maybe in the consumer market, while in every other area, the plan would stipulate how much each company would receive in raw goods and how it would distribute its resulting product, all without the exchange of currency. Essentially, the entire economy outside of the final consumer market would act as a massive meta-company, with each company just handing over its goods between each other according to the plan instead of the mediation of money and prices. These would only remain in the final consumer market, as a simple mechanism to regulate scarcity of consumer goods Or you could eliminatee currency there too and just assign rations to everyone. Needless to say, this proposal is even more radical than regular central planning, and I don't think any country attempted it, for however short a time. And I can't blame them, because it's a high bet with possibly catastrophic costs in case of failure.
Another change of paradigm in order to avoid the communication problem of central planning would be to leave it all to one single megamind. Yes, I'm talking about the cybernetic economic control network that has become a Holla Forums staple. As far as I can see, it solves all the problems above and then some, because they can handle increasingly complex models. You wouldn't be able to tackle overly complex models by throwing more planners at it, but you could do it by throwing more processors at it. In other worlds, the cybernetic model is indefinitely expandable, or at least absurdly more expandable than the planning bureau model, at any rate. I suppose it could be design to work with either price-setting or with commodity quantities, tho I suppose the former would be preferable, simple the main point of the latter would be to reduce calculation complexity, and that's a non-issue when you consider modern supercomputers. And it's important to add, the cybernetic model solves the problem of the law of value by way of simply finding the optimal prices (or amounts) of commodities. In one last note, this model could also free us from the burden of calculating how costly would a social change be, and make us more likely to try them. Remember, socialism was born as a proposal to completely remake society from the ground up, but by the 80s, the USSR was content with being a rather pathetic cargo-cult consumer economy. We simply cannot lose sight of the bigger picture.
tl;dr: cyberplanning is the only choice