The Protein Dial

Background: Mindful Munching with Metrics and especially Gauges and Dials.

The short version of the background here is that there are a bunch of smart things you can do to manage your weight (more to the point: your size) and you should focus on those things more than the number on the scale. Also don’t get stuck in diet mindset. But also, Beeminder is all about bright lines and obsessing about numbers and we think there’s a way to get the best of both worlds. Mindfulness and healthy habits and moderation and all that as the foundation, along with numbers-based guardrails.

Commonly we focus on scale weight as a proxy metric for a fitness goal (but there are other options). We call such a measure of your progress a gauge. You also need a primary metric under your direct control that affects your gauge. We call this a dial. Net calories is a classic dial. Reduce your net calories, see your scale weight trend vaguely down.

One big problem with net calories as a dial (besides how much some people hate tracking them) is that you can create a vicious cycle. The less you eat, the more your body clings to every calorie you do give it. Maybe you find yourself sleeping more, or feeling generally low-energy so you move less.

A related problem with using scale weight as a gauge is that you might lose weight by losing muscle instead of fat. We strongly recommend doing resistance training [1] while trying to lose bodyfat, to avoid that.

But also: protein!

Getting more protein makes everything easier. It helps ensure you don’t lose muscle, of course. And it fills you up, making it easier to run a calorie deficit or avoid a calorie surplus.

So I thought, what if we ignored calories and focused just on protein? Maybe the right things tend to naturally happen when you do that. Instead of reducing net calories when your gauge is in arrears, just up your protein. Is that not abusable, you ask? Would you just eat even more calories in order to hit the protein minimum? Our solution to that is to make your dial metric be, not the absolute amount of protein you consume, but the fraction of your day’s calories that are from protein. (More on this in the “couldn’t you game this” section below.)

Nitty gritty

Alright, the first thing to decide is the bounds (extrema) of the dial. I’ve picked 25-50%. That is, maximally lax is 25% and hyperstrict is 50%. The latter is what hardcore gym types sometimes recommend for what they call a mini cut. (See the related reading below.)

“When your [weight] is not near the bright red line then you just ignore all of this”

Of course, per usual with the Gauge/Dial system, if your gauge is below Beeminder’s maxflux line (think: not near the bright red line) then you just ignore all of this. So even though the “maximally lax” setting of 25% protein is not exactly trivially easy, you don’t even have that restriction when your gauge is far enough in the green.

What we do recommend as general guideline regardless is to aim for at least twice as many grams of protein as your lean mass in kilograms. Like if you weigh 155 pounds and have 30% bodyfat, that’s 49kg of lean body mass, so your target is >98g of protein. This is hotly debated, maybe somewhat less is fine, but all our research sugggests that at worst excess protein is kind of wasted (or can make it hard to get a calorie surplus if your goal is to gain weight) so might as well err high.

At the hyperstrict extreme of 50% it’s actually hard to eat a balanced diet. There aren’t a ton of foods that have more than half their calories from protein. We’ll see an example with chicken breast (77%) in a moment.

Being bound by the protein dial — when you’re in the red on whatever your gauge metric is — means following one rule:

Say your protein dial is set to D, something between 0.25 and 0.5. And say P is the grams of protein you’ve eaten so far today. Then your total calories for the day, C, must always satisfy C ≤ 4P/D.

That’s the bright-line feedback control system part. And of course you let the spreadsheet worry about the math for you, but for the nerds, here’s where that limit comes from:

Your fraction of calories from protein is 4P/C (each gram of protein has 4 calories) and the dial constraint is that this must be at least D. I.e., 4P/C ≥ D. So solve for C and Bob’s your uncle.

“Your total calories for the day, C, must always satisfy C ≤ 4P/D”

For example, if D=1/3 and you have 100g of plain chicken breast for breakfast, that’s 32g of protein and 166 calories. So your calorie cap is 4×32×3 = 384, of which you’ve had 166. So 218 more non-protein calories until you have to get more protein.

Again, this gets hard as the dial approaches D=0.5. But the idea is that your gauge metric will improve with these high-protein days and, with a bit of mindfulness, you’ll keep your gauge in the green and not need these abnormally high protein days often. Of course it’s still great to follow the general guideline every day, to hit your protein minimum. Just that strictness isn’t strictly required when your dial is all green.

Couldn’t you game this by…

Yes, suppose you’re in the red (so the dial rule is in effect) but you get your heart set on a 730-calorie Crumbl cookie, which, sadly, is only 2.7% protein. No problem, even at the hyperstrict D=0.5, if you gulp down (math math math) 896 calories of pure protein powder first, you’re all clear for the Crumbl cookie!

But any system can be gamed and thwarted if you try hard enough. Hopefully that’s an ice cream truck loophole. Also, would you even still want the Crumbl cookie after 896 calories of protein powder?

The Snackwell effect and the so-called Big Fat Lie

Speaking of gaming the crap out of things… Wavy flashback lines to the 1990s, when conventional wisdom was that dietary fat was public health villain number one. Eating fat would clog your arteries and give you a heart attack. But if you take, say, yogurt and add a bunch of sugar then the percentage of calories from fat goes down and you can call it LOW FAT and that sounds healthier. Oy.

One company (or brand of Nabisco I guess) that cynically exploited this confusion was SnackWell’s, which made sugary treats marketed as health food because they were FAT FREE.

The Snackwell effect refers to people eating more of something because it has less of something they think is bad.

I’m particularly amused by the irony here. I grew up not questioning what Gary Taubes dubbed the Big Fat Lie. [2] Then I spent a couple decades shaking my head at how dumb I was to think skim milk was healthier than whole milk. And now here I am, treating skim milk as almost twice as good as whole based on the percentage of calories from protein.

Not that “fat is the devil” has been vindicated. Just that fat is very calorie dense and crowds out protein.

Trying this in real life

I’ve been logging what I eat (and burn) in Google Sheets like so:

It’s pretty slick, I think. I can enter the number of calories directly and ignore the multiplier column (it defaults to 1) as in the 150-calorie protein bar above. If I have a 2nd bar I can put a 2 in the “mult” column. Or I can enter a formula like “=100/113” (shown as 0.8849 above) if a serving has 100 calories and 113 grams, as happens to be the case for 2%-milkfat cottage cheese. Then I can use a kitchen scale to see how many grams I’ve eaten and put that in the multiplier column. Either way, the actual number of calories consumed is shown in yellow.

What about home-cooked food or restaurant food with no nutrition labels? For this I use the Macrofactor app. It lets you take a picture of any food and magicAIally guesses all the nutrition facts. (Or if you’re eating something with a bar code, it knows all the bar codes. You just scan it.)

(ADDED LATER: You people are relentless. Yes, you can drop a stick of butter in your spaghetti sauce and the AI calorie guesser will be none the wiser. The solution to that is, don’t do that. The AI is making a guess based on what it can see and it will sometimes be over and sometimes be under and on average it will be… pretty good, I think? Unless you introduce a systematic bias by making deceptive-looking meals. Again, maybe don’t. And let us link again to our post about ice cream truck loopholes.)

Of course Macrofactor is its own whole system for nutrition, fitness, and weight control. I think it actually has a similar control system deep inside its black box. It shows you your trend weight (if you tell it your weight) and adjusts your calorie and protein recommendations to nudge you towards whatever your goal is. It’s all too opaque for me though, as I suppose it has to be to be usable by normal humans. Still, a Beeminder goal for adhering to its recommendations most days might well work as effectively as this protein dial system.

But back to the protein dial, I also have this gauge/dial dashboard on the same spreadsheet:

As you can see, today my scale weight is slightly above the maxflux line on my Beeminder graph so the target for my dial is 29%, slightly above the minimum (maximum laxity) of 25%. Of the 483 calories I’ve eaten so far, 285 are from protein, so my actual dial is at 59%, far above what it needs to be. I can then enter a hypothetical protein percentage for the rest of the day to see how many more calories I can eat such that my dial doesn’t dip below the 29% target.

If I wanted to reward myself with Crumbl cookies at this point, I’d change that 27.5% to 2.7% and I’d see that final line change to 550 calories. So I could eat like three quarters of a Crumbl cookie and I’d be done. If I, in gross violation of this system, ate the whole 730-calorie cookie, I’d see the final two lines as follows:

• 2.7% protein for remaining calories
• −180 remaining calories eatable

In other words, I exceeded my allowed calories by 180. But the situation is salveagable if from here on I exceed the dial target’s 29% for my remaining calories. Suppose I decide to start eating non-fat Greek yogurt, which has 69% of its calories from protein. I put that 69% in the protein percent cell and now the final two lines show this:

• 69% protein for remaining calories
• 119 remaining calories needed

So 119 calories of yogurt and my dial will hit 29% on the nose. Perfect. For the rest of the day I can eat what I like as long as it’s not below today’s dial target of 29% protein.

The Five Theses of Dials & Gauges

Melanie and I have been debating this endlessly in the weightloss channel of the Beeminder Discord. I gravitate strongly to quantified systems and bright lines; Melanie’s all about flexible rules and addressing one’s underlying psychology. She worries that my approach can engender an unhealthy relationship with food. (Bethany has very gamely experimented with both of our approaches and may have more to say about her own happy medium in a future post.) But Melanie and I managed to converge on the following five principles for getting the best of both worlds — mindfulness and healthy habits along with a feedback control system.

(One more review: you monitor a proxy metric of your health called a gauge and have another metric you control (like protein percentage) called a dial. Adjust the dial, per a formula, to control the gauge. And critically, if you’re far enough in the green on your gauge, you ignore the dial altogether.)

1. Adding dial&gauge guardrails rarely hurts. In theory they can be a crutch, like in that Batman movie where he couldn’t manage to climb out of the pit until he ditched the safety rope and was nice and motivated by a fear of death. Just learn to not let the graph psyche you out rather than ditch the graph.
2. If you get past Thesis 1, adding guardrails is just a backup system. You can and should use mindfulness to avoid ever hitting the guardrails. (And if in doubt, and especially if using scale weight as a gauge, set your red line much higher than you may think you want it. That way you have plenty of breathing room to practice mindfulness without pressure.)
3. If you do hit the guardrails, they help. They do not backfire by putting you in a what-can-I-get-away-with mindset or make you feel guilty for eating “bad” foods or any of that. If you’re prone to those psychological traps, address them separately. The guardrails are a more systematized and distilled version of mindfulness for people who can benefit from bright-line rules.
4. If the rules that kick in when you hit the guardrails are counterproductive in any way, well, we’re all adults. You should be trusted to perceive that counterproductiveness and construct better rules.
5. Or even simpler: dial up (in the informal sense) the strategies Melanie teaches to mindfully minimize bumping into the guardrails in the first place.

Protein in various foods and food-adjacent products

As a handy reference, here are some options for upping your fraction of daily calories that are from protein. To compute these protein percentages, take grams of protein times 4 (usually) and divide by the number of calories. (Sometimes this can get fussy, either due to rounding on the label or because technically some sources of protein have a bit less than 4 digestible calories per gram. This is called Atwater specific factors.)

Thanks to Melanie and Bethany who helped compile these.

• 83-96% for whey protein isolate protein powder. I get pretty sick of this pretty fast.
• 80-91% for liquid egg whites. I haven’t tried this.
• 80% for tilapia.
• 78-90% for canned white tuna packed in water.
• 77% for skinless chicken breast. (50% for skinless chicken thighs.)
• 75% for David protein bars which I think taste better (no worse anyway) than Cliff bars. They’re very processed, of course, but the ingredients seem fine enough. I doubt it’s much less healthy than protein powder.
• 72% for turkey lunch meat.
• 69% for non-fat Greek yogurt.
• 60% for non-fat cottage cheese.
• 47-59% for tofu.
• 59% for salmon.
• 58% for Musashi bars which have grown on me and I think taste at least as good as the David bars. The ingredients also seem fine enough. Both Musashi and the David bars use fake sugar, which I think is fine.
• 53-54% for, what — spinach! Who’d a thunk it?
• 49% for Built Puff bars (brownie batter flavor) which taste, bizarrely, about as good as normal candy bars. Suspicious.
• 43-47% for dry roasted edamame (too dry for me). And also for non-dry-roasted aka soy beans.
• 44% for Quest protein bars (chocolate chip cookie dough flavor).
• 40% for skim milk.
• 34% for a hard-boiled egg.
• 33% for Snickers Hi Protein bars (not palatable IMO).
• 28% for brussel sprouts. These brassicas and leafy greens keep surprising me. Kale, kohlrabi, cauliflower, collard greens — they’re all in here from ~25-32%.
• 27% for lentils.
• 26% for black beans.
• 24-27% for RXBARs.
• 24% for Gatorade protein bars (chocolate caramel flavor).
• 23% for broccoli!
• 23% for cheddar cheese.
• 22% for chickpeas.
• 22% for whole milk (note skim milk all the way up at 40%).
• 20-23% for cabbage.
• 20% for Mealsquares, which are minimally processed.
• 14-22% for whole wheat bread. (We looked up a handful of our local supermarket brands to get a range.)
• 14-17% for Cliff bars (and they don’t even taste good, what the fork).
• 15% for quinoa.
• 14% for oatmeal.

Related Reading

• MacroFactor has a handy article about how much protein to eat.
• Cited in the above is a survey paper about natural bodybuilders suggesting 2-3 grams of protein per kilogram of lean body mass. That would be 150ish grams of protein for me.
• Rob Felty’s protein experiment and results.
• A YouTuber that Melanie of Melzafit recommends, Redefining Strength, recommends getting at least 30% of your calories from protein while on a “diet break” and up to 50% of your calories from protein while doing a “mini cut”.
• As far as I can tell so far, FoodData Central, part of USDA.gov, is an authoritative source for nutritional information.

 

Footnotes

[1] Melanie would like us to emphasize here that you need progressive overload, meaning that you keep increasing weight and reps. In the wise words of Greg LeMond, it never gets easier, you just go faster.

[2] My robominion is pushing up its glasses and saying that technically it was the New York Times Magazine that came up with that title — “What If It’s All Been a Big Fat Lie?” — for Gary Taubes’s article.