Linear Progression: A Myth

[mgil]

(Draft for @LoudMuffin - Not sure if this is succinct enough...)

When we were kids in middle school and high school we were introduced to a simple mathematical equation:

[equation]y = mx + b[/equation]

We learned how to plot this line, find intercepts, and whatever other fun stuff. We learn to use this as a model for things like pricing, distance, and even use it with other linear functions to find simple intersects that optimize the system.

If/when we got to a first course in calculus, we learn to differentiate this equation, finding that the differential is a constant. This tells us that the underlying function never varies in growth.

We all know that training isn't like this. Not at all. But yet many of us have used linear programming and often even recommend it to certain people, like novice trainees. But why?

Well, just like in the applications of a linear model in school, we come to find that this is a simplified view of reality. Like those old questions about driving from point A to B at some constant rate, we know that this doesn't usually happen in reality, especially with other people/traffic (i.e. variables) involved. Yet the linear progression for training exists. That's because it's an approximation deemed suitable for beginners. Beginners/novices have so many variables at play that trying to figure them all out upfront is usually a waste of resources. Start them off with an easy weight, and suggest they add a little weight each session until things stall. It's easy! However, it has a short shelf life.

Novice trainees will acquire skills at different rates. In reality, none of the movements of the slower barbell lifts is overly complex, but there is a spectrum of skill acquisition. Those who acquire proficiency of skill quickly will feel stronger faster. But this doesn't necessarily correlate with the trainee being prone to hypertrophy or maximal display of strength. These could very well be independent variables. The opposite can happen also: someone very prone to getting strong is simply awkward with moving. This could delay progress. Nonetheless, each novice will likely see a somewhat different length of effective linear progression, simply because the model is wrong. And that's okay.

What is being seen now is a separation of variables with respect to training. This is because the data set of people lifting with barbells has become sufficiently large to tease out responses.

The data seem to say that skill acquisition in a lift is likely a short term event in comparison. Sure, form tweaks can be found and made, and those likely create a fairly short-lived response. Cycling accessories is useful to keep the mind fresh and engaged. Even this is an individualistic response because some people are better focused than others.

Hypertrophy is the long-term adaptation most people look to make. This is a slow adaptation and requires increasing training volume of sufficent but recoverable volume to acquire. Data coming available shows that this can be done at fairly submaximal (~50-70%) levels provided the dosing is large enough. Even more interesting is that the length of the sets doesn't seem to matter as much as total reps per session. In other words, working light weights for many sets of few reps can allow for a lot of dose without the per session fatigue cost and deviation in form.

Maximal display of strength is something of interest, but likely dependent on other factors. However, it needs to be trained with some frequency in order to keep the skill fresh, but without impairing training consistency. For most trainees, these displays of maximal strength need to be spaced further and further apart. This isn't a new concept.

While the LP is touted as simple and effective, it's merely effective at being wrong and pushing the trainee to failure modes in the quickest order. Remember, the slope of that function is constant. Some will say, "Of course, we know that, and strength gain is logarithmic/asymptotic!" I'd argue that those are overly simplistic as well. There's likely variance over several windows of time with regards to strength gains that these models would overshoot. In reality, we see people hit a failure mode in lifts even with fractional plates and microloading. When you've shifted training goals to the right of the expected response, probability of success will surely fall. Moreover, the framework most training is built upon that of constantly moving the (e)1RM upwards.

Yet, when you shift training goals to the left (below) the expected response, probability of success increases, but at a risk of hypertrophy. To offset that, you'd simply increase the volume of training. This is a common theme detailed in Data Driven Strength publications and is also found in developments within RTS/BBM programming.

Addiing to this are other external factors like other sources of physical activity, nutrition, rest, stress, motivation, injury, etc. and we know that "simple" programming is ill-equipped to provide an effective training model. Relying on feedback in the loop (e.g. RPE, RIR), fatigue between sets, session video, and other data becomes needed to effectively train.

Overall, LP "works" as in it provides a simple approach to training that has adequate goals and sufficent skill building to bring a trainee to the failure point (maximum display of strength in the situation at hand) in a planned manner. However, none of this makes it the "best" method or even a "correct" method for training. Rather, like most linear models applied to complex systems, it's easy to implement, but undeniably imprecise and wholly inaccurate.

[mbasic]

If/when we got to a first course in calculus,

sic burn btw

[augeleven]

Please submit to SS for review.
Triple dog dare.

[jwilson625]

If/when we got to a first course in calculus,

sic burn btw

what about when i get to my sixth first course in calculus?

[Brad]

Saw this and couldn't resist linking

[mgil]

Saw this and couldn't resist linking

Thanks for posting this!

Mike has an interesting take on this as well. It’s still a bit simplified in some ways, and that one step back to go multiple steps forward is kinda janky. There are ways to not “take a step back” but rather shift training goals to adjust recovery.

Regardless, it’s good that people like Mike are making sure the training public understands that these adaptations aren’t linear and are possibly separable in some cases. The linear model is just a simple model to pull through a scattershot of data.

[mettkeks]

Saw this and couldn't resist linking

Mike has an interesting take on this as well. It’s still a bit simplified in some ways, and that one step back to go multiple steps forward is kinda janky. There are ways to not “take a step back” but rather shift training goals to adjust recovery.

Any kind of undulation has a "step back" built into it, it would be linear if it hadn't?

[mgil]

Saw this and couldn't resist linking

Mike has an interesting take on this as well. It’s still a bit simplified in some ways, and that one step back to go multiple steps forward is kinda janky. There are ways to not “take a step back” but rather shift training goals to adjust recovery.

Any kind of undulation has a "step back" built into it, it would be linear if it hadn't?

The “step back” often has an implication of simply dropping weight and/or volume and simply ramping back up again.

A DUP is planned, so you’ve got various goals, but all are somewhat oriented towards a linear progression in some way.

What can happen is simply shifting a goal in terms of lifts, skills, hypertrophy, cardio, etc to support future training goals. I wouldn’t label that a “step back” per se.

[mettkeks]

The “step back” often has an implication of simply dropping weight and/or volume and simply ramping back up again.

A DUP is planned, so you’ve got various goals, but all are somewhat oriented towards a linear progression in some way.

What can happen is simply shifting a goal in terms of lifts, skills, hypertrophy, cardio, etc to support future training goals. I wouldn’t label that a “step back” per se.

So, semantics? (I don't disagree, btw, I just see this from a different perspective)

You can't keep going when you hit a peak, even in the realms of DUP. All you do is alter the timescale and the perceived magnitude of these effects. The principles have to stay the same whether you assign days or weeks to your microcycles.

"Dropping weight/volume and simply ramping back up" is a very viable way of doing things. It has been since linear periodisation has been around. It only became the wrong thing to do when 3x5 was all we did in a microcycle. That was when we did too little volume AND added too much fatigue at the same time.

[mgil]

It might be semantics, but Mike’s presentation of the method was vague at best. Also, depending on the trainee, you still might have motivational issues, since you’re doing a “regression” while simultaneously expecting gains at or above what was to be expected if the lifter stayed “LP”.

[mettkeks]

It might be semantics, but Mike’s presentation of the method was vague at best. Also, depending on the trainee, you still might have motivational issues, since you’re doing a “regression” while simultaneously expecting gains at or above what was to be expected if the lifter stayed “LP”.

Not starting your next meso @10 is now regression? Or doing a deload week before you start a new meso with 5lb more on the bar?

[mgil]

It might be semantics, but Mike’s presentation of the method was vague at best. Also, depending on the trainee, you still might have motivational issues, since you’re doing a “regression” while simultaneously expecting gains at or above what was to be expected if the lifter stayed “LP”.

Not starting your next meso @10 is now regression? Or doing a deload week before you start a new meso with 5lb more on the bar?

It’s in quotes because of the context it’s being framed in. As in the trainee might feel some sense of defeat doing a months worth of training at a 10% reduction compared to the prior month. Especially if the step back doesn’t realize the gains expected.

Like I said, in Mike’s terms, he’s saying “take a step back” for a month and maybe you’ll get two, three, or four steps forward the next month. RTS does similar stuff with pivot blocks or emerging strategies but the goal is of a different mindset. Yeah, it’s semantics, but sometimes that stuff matters.

[mettkeks]

It might be semantics, but Mike’s presentation of the method was vague at best. Also, depending on the trainee, you still might have motivational issues, since you’re doing a “regression” while simultaneously expecting gains at or above what was to be expected if the lifter stayed “LP”.

Not starting your next meso @10 is now regression? Or doing a deload week before you start a new meso with 5lb more on the bar?

It’s in quotes because of the context it’s being framed in. As in the trainee might feel some sense of defeat doing a months worth of training at a 10% reduction compared to the prior month. Especially if the step back doesn’t realize the gains expected.

Like I said, in Mike’s terms, he’s saying “take a step back” for a month and maybe you’ll get two, three, or four steps forward the next month. RTS does similar stuff with pivot blocks or emerging strategies but the goal is of a different mindset. Yeah, it’s semantics, but sometimes that stuff matters.

He explains what he means very well from 7:45 onward and gives specific scenarios. I don't think there's anything as controversial as you make it sound in there.

[mgil]

@mettkeks, you mean at 9:00 forward...

When I first listened to this, while (honestly being distracted by other stuff, I didn’t get to or simply didn’t hear his explanations that follow the 9:00 block. I also didn’t see the slide where he has the examples detailed.

So on a rewatch things make more sense. I still don’t know if I’d call a shift in goals, especially something like a fat loss phase to go into a gain cycle with a better BF ratio, “a step back”.

[fibula]

Linear progression worked well for me to fix some of my screwups following faddish bodybuilding programs that did less for me. When it does work (and maybe I'm just unique), I don't think that it works for the reasons that Rip & co. think that it does, and I also don't think that it fails for the reasons being critiqued here.

My own mental model is that a lot of the strength gains from "The Program" are systemic. If you are a young untrained male, doing a large quantity of heavy squats will boost your testosterone (or something) and cause muscle growth everywhere. (My fucking hair started to fall out when I did it the first time.) And young males can squat a lot more than they realize when they are forced to. I did find that the technique advice from Rip is mostly value negative, and gets worse over the years.

But the dirty secret about the programming -- one that I didn't realize at first -- is that "linear progression" has nothing to do with trying to match your lifting to the correct percentage of potential or whatever. As mgil points out, you could get better at accuracy of percentage of 1RPM potential using feedback, RPE or something else. But I expect that you get little real training benefit from this accuracy. The systemic squat effect just isn't modified too much by better %1RPM accuracy. Food/sleep/outside-activity matter more. And the added training complexity takes more time (and can go wrong).

One thing that linear progression did well (at least it for me) was to fix my broken/untrained internal feedback mechanism that told me what weights are too heavy and what I couldn't do. Micro-loading, etc., has absolutely no training or physiological effect, but it's simple to implement and psychologically effective.

The volume/intensity thing is real, although when I've personally hit actual overtraining -- the real kind -- volume and cardio and low calories and especially lack of sleep have been the enemy. I really wish that I knew more about what's possible and what's dangerous here, especially when I try to build better maintenance programs (which I've never been as successful with as I would like). I've read the sources that DDS/RTS/BBM use for their numbers, and I have often been less-than-impressed by how much mileage that those guys have gotten out of various studies that I'm much less confident about. A lot of it is just better marketed bro science. I remain partial Rip's badly marketed repackaging of Starr in comparison.

[mgil]

@fibula, I appreciate the feedback.

I do think there is value in simple programming. Especially simple programming over compound movements with relatively long ROM. And yes, the simplicity of the LP allows for reduction in psychological noise, so to speak.

Regarding Starr, when I went digging through some of his stuff again recently, the brilliance of ramping 5s became obvious for time constraints with good physical responders. Literally - sets of 5 at 60%, 70%, up to 100%. But the 100% is just of the planned lift, and you can increment that as needed. It’s every rep, warmups included, packaged for everyone with no room for wondering what to do for the entire session. It’s all plainly laid out.

[fibula]

Starr's Strongest Shall Survive is wonderful. I wish I had a physical copy. My understanding, from pg. 59, is that you increment the planned lift every heavy day (ie., once a week). He starts out with 10lb/week increment for presses, and 20lb/week increment for squat, until it doesn't go up like that anymore. Basically linear progression, as far as I can tell.

Reading SSS, the utility of the HLM program is really obvious to me in Starr, where football training happens during the week along with weight training, in a way that never makes sense to me as a simple powerlifting recovery-cycle template. The point would be that during football season you can time your heavy days to be the day in the week where you are maximally recovered from football (ie., Monday).

But other than HLM, Starr's program differs from Rip's with warmup + 3x5 instead of 60%/70%/80%/90%/100%. I'd give points to Rip on the 3x5. There isn't going to be an appreciable training difference -- it's basically the same workout -- and it's a lot simpler to do the math with 3x5 after you hand-wave the less important warmups.

[asdf]

I'd give points to Rip on the 3x5. There isn't going to be an appreciable training difference -- it's basically the same workout -- and it's a lot simpler to do the math with 3x5 after you hand-wave the less important warmups.

Starr's ascending sets of five are useful for lifters who are inclined to go too heavy. Even if you grind your top set of five on your heavy day, that's better than grinding two or three sets of five, which seems to be what happens to everyone on SS sooner or later.

[fibula]

I've been trying Starr for the last couple of months. Things I didn't like:

* Way too much fiddling with weights
* Is the last set my "work" set? The last two sets?
* Warm-ups are a bit heavy
* It's easy to screw up rest times

These last two have the combined effect of making the last sets a grind. Things I liked:

* Heavy/Light/Medium is a great schedule
* One and done for the heaviest sets is motivating
* Light and medium days are pretty short, as I can largely eliminate rests. Even heavy day only requires 1-2 more serious rest periods per movement.