In Coaching, CrossFit, Reference, Videos

April 18, 2010

Video Article

CrossFit is about constantly varied training, but that doesn’t mean the training is random. While it’s true that we train for the unknown and the unknowable, programming is more than pulling random workouts out of a hopper—though you can certainly experiment with that model for short periods. In this clip, shot at a Coaches Prep Course in late January 2010, CrossFit Co-Director of Training Dave Castro talks about the art behind constantly varied programming.

One of the main characteristics of constant variation is the use of very different loads, time domains and movements. Sometimes athletes are performing short, heavy workouts like 5 sets of single deadlifts. Other times they’re doing a long, heavy workout such as Linda. Cindy is long and light, while Fran is usually short and relatively light. A 2K row and a max snatch both involve pulling, but one is about endurance and stamina, while the other is about strength and speed. Athletes adapt to such a diverse program because it is broad and ignores no aspect of fitness.

Constant variation is one of the foundations of the CrossFit program, and by allowing that principle to influence your programming, you will be putting your athletes on the path to high levels of general physical preparedness. As always, the magic is in the movements themselves. Add in large doses of intensity, and you might even have some true firebreathers on your hands.

The Coaches Prep Course builds upon the Level 1 Certification and is designed for coaches looking to take their training to the next level.

6min 28sec

Additional reading: What Is CrossFit? by Greg Glassman, published March 1, 2004.

Free Download


24 Comments on “Constantly Varied Programming”


wrote …

My takeaways (maybe call it personal notes for the training notebook):

Why we do what we do: General Physical Preparedness (GPP)

How to get there:
Constantly Varried (CV)
Functional Movements (FM) excuted at
High Intensity (HI)

“Variance is not doing, what a lot of people think CrossFit is, and do whatever the fuck they want to do everyday, and have no plan.”

“The magic’s in the movements; the art is in the programming.” (Maybe this is the sand that fills everything in)

Great stuff. Thank you for educating us daily.


wrote …

After watching the video twice I still am unclear on one aspect, and I invite anyone to clear it up.

If variance is NOT random. What is variance?

Dave suggest that variance is changing the load, time(ie intensity), and duration of a work out. But if those three are randomly changed wouldn't that in fact be a varied workout?

Perhaps the "hopper" programmer is only ineffective because there is a limited number of WODs in the hopper? If a hypothetical hopper existed that contained every motion that CF has implemented would that be an effective programming method?

Any comments you could provide on your take on variance would be appreciated and will certainly make my programming better.


Tommy Marshman wrote …

Justin I think the issue is you are using variance and random as interchangeable which in this case they are not.

You can have variety both random and controlled.

I think the idea is to control the variety of movements to get the best effect from the training and programming.

Where as random variance may, at best, only achieve the same reults by chance if at all.

That's my take on it. I think the programming is what seperates a trainer from a coach really and it's something I'm trying to get my head around. For me it's one thing to understand the individual movements and their requirements but to fit them all together in the puzzle is the art in the science.

but I digress.


Frank DiMeo wrote …

Thanks Dave!
Very helpful insights!


wrote …


In anything truly random, you end up with what we call clusters or groupings. Variance on the other hand is a measure of the departure from the normal or the average. In context to this video, variance is a controlled, measured departure from the normal to elicit a desired response.

At least that's my two cents.



replied to comment from Dustin Standel


Sure, truly random does mean you can clusters, but the mainpage throws in clusters every know and then (randomly). Your post seems to imply that A) clusters are bad and B) clusters are unavoidable. As a coach, you can go random avoid clusters. The number of workout options is seemingly endless.

I don't understand Dave Castro's point. How does constantly varied not imply random? It's not random in the sense that you ignore the other two components of the sentence "functional movements" and "high intensity". Once you stop being completely random, aren't you just making it consistently varied and changing the goal like crossfit football/CFSB or crossfit endurance. Those programs have functional moves and high intensity but lack the 3rd part.

The hopper model over long enough time should produce the most balanced GPP.


wrote …

Ew, I wrote know and then. Shameful. In fact, terrible grammar all over the place.


replied to comment from Matt Solomon

"How does constantly varied not imply random?"

Because if the person in charge decides that tomorrow's workout is going to have deadlifts and ring dips in it, that's not random, it's deliberate.

Programming can be varied merely by ensuring that no single pattern is repeated. Randomness is not required to achieve that -- especially with a large enough set of movements, durations, sets, reps, rests and loads to choose from.

"The hopper model over long enough time should produce the most balanced GPP."

I think this notion puts too much focus on the "general" element of GPP and not enough on the "preparedness" element. If you go 12 months between deadlifts, you simply will not be as well prepared for it as someone who deadlifts regularly. Same thing with 5k runs. Adaptations to stimuli require a certain amount of time (i.e. repetition) to take hold, and they fade within a certain time as well.

So, to truly be prepared to perform well at a given task requires a certain amount of focused, planned practice of that task. The question is how well you want to perform at that task, what is required to achieve that level of performance, and what you have to sacrifice in training for other standards of performance.

CrossFit programming includes a larger set of variables than programs that train for SPP, but there is still a certain amount of specificity involved. The question that everyone wants to answer is just how much generality and variance, and just how much specificity and regularity, to include in their programming. And that, of course, depends on their goals.


wrote …

Saying deliberate is just changing the semantics.

Going for a 5k run should not improve your max deadlift and vice versa. Have ten WODs a year that are 5k runs, means you can't do 9 different things. Repeating workouts is good to prove that your capacity is improving, but does it really help all domains? The unknown and unknowable? If I NEVER repeat a workout, I know that I can still improve my fitness.

Look, you can do Fran every 3 months, or the next time it comes up you could 22-14-7-2 reps instead of the usual 21-15-9. Yes, I chose that, but I randomly typed the numbers. Attacking that four round set with as much intensity as the 21-15-9 will undeniably tax your body and improve your capacity.

I don't have to do heavy deadlift singles repeatedly, but I won't necessarily lose that strength if I'm also doing heavy squats or heavy OHS or sumodeadlifts or heavy sumo deadlift high pulls or snatch grip deadlifts (with elevated feet). Or doing normal deadlifts at 3x3.


replied to comment from Matt Solomon

"Saying deliberate is just changing the semantics."

No, it's changing the meaning. Random programming is uncontrolled. Dave Castro controls -- deliberately -- the mainsite programming.

"If I NEVER repeat a workout, I know that I can still improve my fitness."

Perhaps... but without repeating specific events, you can't quantify that improvement.

"Repeating workouts is good to prove that your capacity is improving, but does it really help all domains?"

Increased work capacaity across BROAD time and modal domains is NOT the same thing as increased work capacity across ALL time and modal domains. One is the stated goal of CrossFit, the other is your invention. Good luck with your 17-day continuous run and your 8,000-pound deadlift! Or with your exhaustive consecutive series of Frans done with different permutations of 45 reps!

The training parameters of CrossFit, fortunately, are finite. They are general only RELATIVE to other programs. Programming within those parameters allows for a great deal of variance, but succeeding at consistently increasing work capacity across broad time and modal domains also demands an intelligent, artful programming design that enables the simultaneous creation and maintenance of numerous adaptations. Truly random programming simply can not accomplish that.


wrote …

I think everyone is reading to much into the words and into the programming,they are several ways to skin the cat, but the end result will be what you want it to be. You use the programming how you want it to fit you and what results you want. Do you want to be able to say you can do every movement in CF, or do you want to get stronger, faster, lose weight, or whatever your goal is. Variance, random, hopper, and all this other crap words, may seem cool to use, but the bottom line is understand the program how it fits you, your athletes, or members. get what you want out of it, and run with it. We can come from all different angles or ways to skin the cat, but at the end its gonna be skinned one way or another: Soooo.... set goals , make a plan, be ready to adjust, and Let's do it.

these are my two cents, just thought I share some insights. I know not everyone will agree, but that's the beauty of these blogs, we learn from each other.



wrote …

Sorry but aren't there any other people working in CFJ other than Dave and Tony ?


wrote …

Two points:

Random workouts
No one programs using purely random methods. A purely random method would include: select a random modality (for example pullups) now select a random number of reps (10,392 for example). I don't think any one would program 10,392 pullups for a workout.

When people say random, they mean something different. I propose that they mean "select a random workout from the subset of reasonable workouts".

How would you tell the difference?
I am going to play a bit of devils advocate here...
Lets assume that workouts are *not* randomly selected.

Now imagine a box that programs their workouts via a random method (the hopper for example). If 'random' and 'variance' are truly different then I (or someone smarter than me) should be able to look at both the workout programs and select the 'random' program. I don't think I could tell the difference.

I would be interested if someone had a real algorithm that could tell the difference. If no one can provide a reasonable algorithm to identify the random workout then is their really a functional difference?


wrote …

Hoping it get better...I think I've seen similar material done much clearer and more concise...maybe in India...


wrote …

that's strange... some comments were deleted. and they weren't disrespectful or anything like that. i wonder what happened.


wrote …

As someone who programs for a box, our workouts are constantly varied but definitely not random. We have in mind a clear idea of what we want to hit, when we want to hit it and for how long we want to hit it.

IE, Say we know we want a 20 min, lightish WOD that incorporates gymnastics and met con. We'll so something like, AMRAP/20mins, run 200m, 10 Pull Ups, 5 HSPU's.

The next day we may have Back Squats of 3 ascending sets of 5 (we run a strength bias program) followed by a short, heavish WOD of less than 10 mins. We might do 21-15-9 for time of 32kg KB Swings and Ring Dips.

The WOD's are planned taking into account overall volume for the cycle we're in, in addition to what we've already done for the week, what we plan to do for the week, the identified training needs of out athletes and logistics such as weather and equipment. The next day we will do a different WOD and the next day another different WOD. Hence, the programming is constantly varied, but very much planned. We have benchmarks (our own as well as CrossFit girls) that we return to every so often to gauge progress against those benchmarks.

Constanly varied and random and entirely different things. Not just semantics. Pulling random shit out of a hopper and throwing it all together is ok every now and again as a WOD, buyt if you base your programming on that, good luck. You'll need it.

Hope some of that helps.

The Cell


wrote …

The (although at first glance it doesn't seem like it) clearly defined goal of "Increased work capacity across broad time and modal domains" forbids truly random programming. Even more so since it implies progression and therefore requires measurement! Add to that K-Star's insight that "the only thing not perfect about CrossFit is the athlete" and you already have the keys to programming: as in goals to meet and restrictions to overcome.


wrote …

Great work Dave.
I appreciate the knowledge, very clear and easy to understand.
Wish I was in that class. Will get my ass to one soon!



wrote …

Let's.look at some of the girls:
Diane: deadlift+hspu
Elizabeth: cleans+ring dips
Fran thrustters+pullups

if you stick these into a hopper, some day you'll get hspu+dips+thrusters
then another you'll get deadlifts clean and pullups. And there might be rowing in there too.

Noone would be expected to maintain intensity on such workouts.

Likewise, random programming isn't going ensure local weaknesses get targetted more than their fair share.


wrote …

Some definitions of 'random' (source:

1. Unpredictable
2. Assorted; undistinguished.
3. (pejorative) Frivolous; unproductive; undirected.
4. Incoherent or inelegant; poorly chosen; not well organized.
5. In no particular order
6. Arbitrary.

Some definitions of 'varied':

1. To change the aspect of; to alter in form, appearance, substance, position, or the like; to make different by a partial change; to modify; as, to vary the properties, proportions, or nature of a thing
2. To change to something else; to transmute; to exchange; to alternate.
3. To make of different kinds; to make different from one another; to diversify; to variegate.

I vote for varied.

The way I have always understood it is that CrossFit using constantly varied programming to increase an individual's GPP to be ready for totally random tasks or tests. So the programming should be constantly varied (altered and different in load, duration, intensity, type of movement, etc.) to prepare for the random.

I think random can and does work in the short term, but it also can hurt your overall GPP if you are a slave to it. If heavy back squats come up randomly and my squat was disproportionately stronger compared to my body weight than my ______ (dead lift, clean, overhead squat, front squat, etc.) was, than I might use the same rep scheme, load and intensity but insert a different movement. Still varied, but tailored to my weakness vs. my strength. Or, if Met-con is your weakness you could make it a couplet with burpees or lower the weight slightly and keep the same weight but shorten your rest between sets to one minute. I think that was Dave's point of constantly varied vs. random.

This is just my $.02. I am no expert, just another CrossFitter.


wrote …

Hi to everyone. English is not my native language so I apologize for my grammar mistakes. I hope I will be understood.

Everybody talks about programing only in the terms of reps,loads and etc. but I want to know how did you come up with those numbers? Why 7? Why not 5? Why 21 and not 41? When you change the numbers you change the purpose of the workout. What is the purpose of Fran? Strength? Stamina? Flexibility? You cannot say that Fran or any other WOD develops all of those abilities. I understand that you can influence those abilities to some degree because human body is a unit and all the abilities are working while we execute the workout. But if you want to increase strength than you have to lift heavy loads and you have to rest awhile because your CNS simply cannot handle it. The same and even more important is with power. Now I do understand that high intensity also means great loads. What I want to say is this: crossfitters lift amazing loads for non professional athletes. Their deadlifts and OL are awsome. If you want to increase your PR on the deadlift you need some kind of programming that will get you there. For example: Monday= 80% 1 rep Max, Tuesday: 90%...and than you have to choose the exercises you will do and how many there will be, in which order you will do them and for how long will your strength or power cycle last, how long are your rest periods, how much time between workouts during the day or how many days between heavy workouts, when do you plan for PR workouts...that is programming. Knowing what you wish to accomplish, determening the means by which you will do it, choosing the way to do it and execute have to have the preparation fase, the competition fase and an off season fase. I am sorry but THAT IS PROGRAMMING! I would be seriously surprised if the crossfitters attending the Games do not use that kind of programming if they wish their numbers to grow! If you only do Fran with the same load than that is the load that you will be able to lift. Your time will change and you will have more stamina but your strength is going nowhere.

So, anyway that is what I am saying. If someone can point out some video clip or a book or something...anything that talks about those issues I would be gratefull.

And by the way I still think that crossfit is an amazing program and has a great deal of cool stuff to offer but I also think that some things are unsaid.

- Sasha - (Croatia)


wrote …


Don't forget that crossfit is a gpp (general physical preparedness) and is aimed at building a strong foundation for whatever you want to delve into whether it is team sports, endurance events, or strength events. Obviously if you want to get better at a specific task you will need to incorporate that task into your training but by having a strong foundation of fitness in the 10 areas you will be better suited to attack that task while reducing your chance of injury, increasing recovery time (increased aerobic fitness = improved recovery in other energy systems).

If you take it from a periodization stand point even microcycles/mesocycles should incorporate some form of general training before delving into sports specific training (read the science of sports training). My goals in crossfit are for increased fitness, increased knowledge and the ability to coach, and finally performance. That is why I love pr's but I don't let it ruin my day if I don't acheive one. Sometimes you have to look at it as "did you achieve the training effect" - i.e. strength day, short metcon, skill, etc. If performance is your goal than obviously you will need to alter your plan of attack to include your weaknesses, that is why I think if you look at those that train for the games you will see them go away from mainsite programming to more individualized plans of strategy that may involve a strength bias like you mention or a gymnastics portion or skill work...

The beauty of crossfit is that a crossfit athlete will be able to outlift a runner and outrun a lifter.


replied to comment from Brian Hassler

thanx for the book advice. I will definetly read it.


wrote …

I might be beating a dead horse and reopening up a can of worms. I'm just new to Crossfit and am learning how best to program for myself and for my athletes/clients.

I agree 100% that constantly varied does not mean random. I think that this is pretty much a general feeling in the wider community these days. Most comments on this video are dated 2010 and it's over 2 yrs since the last comment.

In Supertraining, Siff & Verkhoshansky (2nd Ed, p33) state that there are 5 interdependent processes that an athlete must achieve:

1) Physical preparation (general physical prep, and specific physical prep)
2) Learning of motor skills
3) Psychological preparation
4) Physical and psychological restoration
5) Appropriate nutrition

Generally speaking, I'm sure you'd all agree, that most great Crossfit boxes are starting to integrate these processes into their programs. For sure, Crossfit is and always will be the best allround GPP program out there, but why?

It is constantly varied, high intensity, functional movement.

So I come back to my point. Varied doesn't mean random.

I contend that a program can consist of several blocks or units of a program and have workouts or components of workouts repeated within those blocks.

For example:

Week 1-2 - 6 day/week program, repeated in the second week to assess benchmarks achieved in week 1.
Week 3-4 - A completely different 6 day/week program, repeated as above
...and so on.

This is still constantly varied in my opinion. And there are a plethora of other alternatives to this.

For example:

Week 1 - a 6 day/week program 1
Week 2 - a 6 day/week program 2
Week 3 - a 6 day/week program 3
Week 6 - same as 1
Week 7 - same as 2
Week 8 - same as 3

This is still varied.

Put your creative caps on and think about it for a while and you can come up with 26 weeks of programming easily enough.

Benchmark WODs and other workout elements (such as back squat 1RM, RE pushups, etc) should be passively assessed constantly (ie tracked) so that the resultant athlete progress can be monitored. We want improvement in general terms - hence we aim for GPP.

I feel, that even though we are aiming to achieve general physical preparedness, there are going to be weak links in fitness capabilities. For example, an athlete with a powerlifting background will be weaker in general endurance than another athlete more exposed to cardiorespiratory training. Is it wrong to assume that a good GPP/Crossfit program would include a more skewed focus towards endurance training (for this athlete) if this powerlifting client wishes to develop true general physical preparation?

Of course that is fine!

Just because there is a focus towards one particular fitness modality doesn't mean GPP can't be accomplished, in fact it is probably optimised!

These are just thoughts, feel free to add/criticise and contend. Feel free to teach me and everyone else :) I'm here to learn :)

Basically, my point is, I think Dave Castro scratched the surface of what constantly varied truly means. There are a plethora of ways to achieve varied training. And it is always important to understand that GPP is only a part of athlete preparation. However, GPP training is absolutely essential as a foundation to achieve true fitness.

I look forward to doing my Coach's Prep course in due time to build on my knowledge :)


Leave a comment

Comments (You may use HTML tags for style)