This seems to apply to plans/intents more than it does for advice per se. Perhaps you model a consideration of action as self-advice? Or you're talking about implementation of advice? I tend to think of evaluation of advice in terms of "what goals does the advisor think I have?" and "what does the advisor know about those goals or my situation that I don't (or haven't noticed)?"
I think in terms of your framing, I'd use different dimensions than underfit/overfit. I'd focus on completeness and specificity. "work hard" is almost always correct. It's not very complete, without some description of what kind of work, and how to work hard. It's not very specific, since it doesn't specify how to change from how hard you're already working.
Perhaps I'm overenthusiastic about this post since the topic of advice has been somewhat of a fixation of mine recently but the points about correct advice being specific and the overfitted and underfitted distinction strike me as very useful for querying if what has worked for others will translate over to one's current situation.
One thing I might quibble with is the phrase "correct solution" and even then it's not really a quibble, because a correct solution is any solution that has more (relevant) benefits than costs. However there are degrees of correctness, this of course overlaps with overfitting and underfitting. There are often multiple correct solutions to any given problem. Some are better fitted than others. But for advice to be correct all it needs to do is "work" - that is the advisee needs to be better off for having followed the advice than if they didn't.
An analogy that comes to mind is I often use a swiss army knife to cut things like masking-tape or strings from clothes, whereas a pair of scissors might be more effective and easier. The knife is still a correct solution as it does a satisfactory job. Scissors might be "more correct".
A flow on thought from this how correct solutions can be arrived at. The more expeditious method is to share the situation with a expert in a relevant domain who will then draw upon their expertise, understanding, and modelling of the world/domain to offer advice, or the slower method of arriving at a correct solution one's self. Both getting expert advice and discovering it without help as you point out require specific knowledge of the problem.
Repeated experience has taught me vague or underfitted advice is often a symptom of vague descriptions of the problem by the advisee.
The interesting thing is that even in the absence of an expert, the more specific someone can get about defining their problem - the easier it is to find correct solutions.
I have a bad habit of being vague and underfitted: I'll ask myself a question like "why can't I network?" to which I might get an answer like "because you don't promote yourself enough". Whereas if I focused on a specific problem and gave situational knowledge like "I've looked on Facebook and Meetup but cannot find any groups or events which suggest themselves as being appropriate contexts for finding...etc." the specificity drastically improves the fit of advice and the chance of arriving at a correct solution. Whether the solution concerns which search terms to use. A change in expectation or belief about the idea of appropriateness. Looking on a different website. etc. etc.
Most advice is bad. Whether it’s personal advice or business advice, you probably shouldn’t follow every piece you hear.
While I recognize the irony of beginning this with a piece of my own advice, there’s a specific explanation as to why I’ve done this. Being selective about the advice you follow is perhaps the only universally appropriate advice in business. People and businesses want help solving their unique problems, yet most advice is not appropriately specific to their situation. Such advice stems from the advisor’s circumstance, rather than your own.
To find appropriate solutions to problems, specific knowledge has to be created around the framework of your specific situation. This process is what generates a correct solution for your problem, and your problem only. To pass it along would be to participate in the cycle of bad advice-giving.
To see why the cycle of bad advice is so common, it’s helpful to walk through the trajectory of a typical solution to any given problem:
No explanation → heuristically correct (underfit) → truly correct → heuristically correct (overfit).
Imagine you want to grow an online brand. You start without knowing what to do, and despite witnessing or maybe even “risklessly” analyzing other brands, it’s different than acting yourself. You have no explanation.
If you start experimenting, learning, and growing, you’ll move towards being correct (but you won’t be correct just yet). The first pitfall that emerges is being heuristically correct. A heuristic is a mental shortcut for decision-making that abstracts away some complexity of a situation, giving you a simpler (sometimes simplistic) model of your problem. The first failure mode of being heuristically correct is underfitting. This type of business advice is easy to spot: it’s composed of trite, uninsightful ideas:
“Be confident,”
“Post consistently,”
“Work hard.”
This isn’t really “bad” advice, as it won’t nosedive your business by following it. However, it doesn’t give you direction towards creating and capturing value in your specific domain. This is usually the sort of advice you’ll hear at a conference or Q&A, partly because panelists aren’t privy to the nuances of your situation, and partly because they don’t have the time or the will to provide a comprehensive solution. So what you’ll get is platitudes that are generally correct, but leave you disappointed regardless. We all know that working hard is good, but simply driving ahead without a plan might result in a Sisyphus-esque situation. Blockbuster did not fail due to lack of effort, but due to poor alignment with future trends and an inability to shift quickly.
Back to growing your online brand: you’ve gotten lots of pointless, underfitted advice; but you’ve also kept experimenting. And since your brand is targeted at 16-22-year-olds looking to learn a new language, you realized you could get more impressions if you share “crazy” meme content with your owl mascot since this creates high brand virality, which has worked out well for you so far. You implemented a truly correct solution.
Being correct is using the ideal combination of strategy that matches your environment and tactics you’re uniquely suited to execute. You usually earn correct knowledge through rigorous effort, such as with bottom-up data analysis or top-down problem structuring.
Correct advice is specific. It applies to no one else, in no other situation. A great consultant, coach, or mentor gives correct advice, yes, but more than anything, they give advice that is specific. You could often arrive at correct solutions yourself, but a thoughtfully engaged advisor can help supercharge the process of conjecturing ideas and criticizing the bad ones, leading to correct solutions.
However, these advisors are not always thoughtfully engaged and often give bad advice anyway. Why? It’s because they often provide heuristically correct solutions, which are equally as lazy as the underfitted platitudes from before, but are disguised with a misleading facade of specificity. These are overfitted solutions, copied as newfound heuristics that worked in other success stories, and pasted with limited or nonexistent tailoring to your specific situation.
For example, RedBull littered crumpled cans near waste bins to generate buzz for their product. That doesn’t mean you should also scatter your online brand’s merch blindly and hope it works out in a similar way. RedBull succeeded because fledgling CPG energy drink brands need social proof and the illusion of popularity, but unless that applies equally to you, it likely won’t work.
Novel tactics generally only work once—an epistemic monopoly—where a new idea can exhaust the niche it created, and lagging imitators must overcome both the consumers’ habits with the innovator’s implementation and also customers’ fatigue from repetitive copycats. Customers may have been inclined to download the McDonald’s App when it was released because, at the time, it was a new dimension of competition in fast-food. But as competitors scurried to make apps of their own, it’s worth asking if customers have a limit for how many apps they’re willing to download.
Since underfitted ideas don’t get the intellectual glory that overfitted ones do, people building their brands often steer towards overfitting. But overfitted ideas can be just as useless in solving your problems. They can be actively destructive, by sinking time or money into ostensibly surefire plans. A useful gut check then is to ask yourself: are you trying to emulate a specific company or playbook too closely? Is that happening at the expense of trying strategies, checking whether they worked, and iterating?
Overfitted ideas may incidentally solve your problem, but usually not for the reason you think. They usually introduce unseen problems, and often impede the success you could have found without them. Often, when overfitted ideas solve your problem, it just means you got lucky.
Online “experts” entrap you with one of two heuristically correct temptations: providing vastly under-fitted ideas and sounding impressive (“Work hard and make millions!”), or selling you a foolproof plan to scale your business (“Buy my twenty-page PDF and online course!”).
Correct solutions come from doing things yourself or finding someone who will do it alongside you.
Correct is not pre-packaged, and something generalizable is almost never correct in the way you need it to be.
“Correct” that only works when luck strikes is not worth betting your business on.
You wouldn’t risk your future on bad advice. Don’t risk your business on it either.