There is no optimum
A few thoughts on why optimal programming is impossible in fitness and beyond.
I personally like the 5/3/1 system but I’ve seen lots of online chatter about its deficiencies. And if you google search for strength programs, there are hundreds of them. Everyone has a program that’s going to max your gains to the next level. Even if you are just interested in powerlifting, where the goal is to maximize your capacity on squat, bench, and deadlift, there seem to be infinite options out there.
And even the popular options for programming are wildly different from each other. There’s “Starting Strength” where you simply do 3 sets of 5 reps every week and add weight to the bar if you complete the sets. There’s the “Texas Method” where you’ll squat 3 times a week, but with different intensities to target different objectives. There’s the “Bulgarian Method” where you just max out your main lifts six days a week. Please don’t do this one. There’s “The Conjugate Method,” which I also wouldn’t recommend running unless you’ve watched the documentary “Westside Versus The World” and decided that mindset was what you were looking for.
But why are there so many varied approaches? I got stuck this morning in a technical description of what these different programs are all chasing. But I deleted that post and came up with a few unsatisfying but probably true answers. And again, I want to use these answers to poke at the bigger picture of conceptualizing interventions for individuals.
People are selling you a brand.
There’s an undeniable commercial core to the fitness industry. I don’t want to dwell on it. But even the best-intentioned coaches have to make a living, and people who write programs are trying to sell you something at the end of the day. Selling a program requires differentiating yourself from the crowd. But it’s not just money that drives differences.
Lots of things work.
I am very biased as an optimization researcher, but there isn’t a “right” algorithm for maximizing an objective. You can try gradient ascent. You can do stochastic gradient ascent. There is Newton’s method. There are quasi-Newton methods. There is ADAM, the Starting Strength of optimization methods. All of these will likely work in the appropriate context, but each will work differently on different objective functions. What method should you choose for your problem? There’s no good answer! You have to use a bit of prior knowledge and you have to experiment. Steve Wright described optimization methods as implements in a tool chest. Each has a rough guideline, but you never know which will be right for the job until you dig in.
The effectiveness of a program depends on your experience.
This is also sort of obvious, but programs are written to target people with wildly different skill sets. The basics of “Starting Strength” are probably find for the first couple of months of a very novice lifter. But a world class weightlifter can’t add 5 pounds to their clean and jerk every week for perpetuity. That fitness training necessarily needs to be personalized is one of the reasons I think it’s a perfect foil to the “one RCT rules them all” causal inference gorilla.
Measurement is hard.
While it’s clear that “performance” in sports should be easy to quantify by split times or weight lifted, performance varies from day to day. Moreover, because of injury risk and fatigue, we can’t test maximal performance every day. There’s a conventional wisdom of that working at “80%” is most effective most of the time. But a good program needs to assess “what does 80% mean today?” This is done by a variety of measurements, each with its strengths and weaknesses. Some programs calculate weights based on percentages of the most you’ve ever lifted. Some programs calculate weights based on “perceived exertion,” suggesting you stop when you think you have 1 or 2 more reps before you are exhausted. Some programs suggest you do as many repetitions as possible today. All of these can be effective, but you’ll have to design the particulars of a program around which measurement you choose.
There are tradeoffs.
There are multiple competing objectives when you are exercise training. Even if you are just in it for general health, you have to balance many factors, including muscle strength, joint health, and endurance. Again, as an optimization researcher, this resonates with me. Stephen Boyd famously says, “everything is optimization.” For a long time, I agreed with him. But, and this deserves a much longer post, I want to rant at length about how this is wrong. Let me just say here that the multiobjective problem already shows why you can’t optimize everything. As soon as you have two objectives, there is an ambiguity that optimization can’t solve. And hence, getting back to exercise training, part of the variety in programs is explained by their very different goals. When there are trade-offs, there can’t be an optimal program.
These five are some of the answers. I want to keep these in mind as I’ll turn tomorrow to see if these ideas of adaptation, overload, and periodization can apply to anything outside the world of physical fitness.