Applied Nate Silver – It’s a pattern until it’s not

This data and graphical view shows how patterns seem to appear in data and then more data comes in and the pattern is violated which implies the apparent pattern really was never there. I’m reading a book, On Second Thought: Outsmarting Your Mind’s Hard-Wired Habits, by Wray Herbert and it is describing patterns of thought, believed to be hardwired into our minds, which the author labels as ‘heuristics’. And one such heuristic is that our visual processing, which is outstanding at spotting patterns (so we can distinguish the movement of grass as the predator about to jump us vs just a puff of wind), extends to abstract tasks as well, like looking at data. Nate describes how sometimes we read too much into data that looks like a pattern when it isn’t. In a defense of his correct election predictions he attributes the Romnoid’s error in believing the data supported their notion they were winning as this tendency.

Well now I’m guilty of this. I’ve been upset (that’s mild, panicked is more like it) that I’m stuck on a weight plateau. So after an extreme effect I finally saw a drop and a drop that was too large to be explained by my actions, in short, yet another piece of noise in what I’m increasingly finding is hugely noisy data. So naturally I looked for patterns and found it. So I predicted that no matter what I did the next day my previous day’s anomalous loss would be wiped out and it was only a question of by how much. So here’s the graph:


Note the three points marked with yellow: big drops (too large to be real) with each of them followed by an increase in the next day. From a fundamental point of view it is easy to say the big drop was a statistical fluke and so the reversal is not so bad. But it’s not so easy to rationalize this to my emotional side – I love the big drops and have the big gains even though I know the drop was fake. I just hate to give back any gains, even if not real.

So look at the data highlighted in green. Once again the big drop (a drop was expected but this is too much, not believable from any reality or science) and so I was prepared to see the same pattern as marked in yellow, that the next point would pop back up again. In fact when the anomalously low point is followed by a higher point it’s by over 2lbs and I was cringing at that idea since it means the two-day change would be tiny relative to the trendline.

So naturally I was very surprised when that didn’t happen this time. Not only did I not reverse the low point I actually “beat” it. This was particularly amazing to me in that the very first scale value is that lowest blue dot, the most anomalous possible.

So now what do I do – believe this? Or is the pattern that it will take two days to bounce back up again? I’ve been disappointed so often, after all the mental effort this weight loss requires, at any uptick, so naturally I’ve fallen into “pessimism” about upticks (predicting them always) as a defense (“oh, it’s just the statistical fluke, don’t worry about it, I’m still on long-term trend”). Pessimism is a rationalization to defend against disappointment; either my pessimistic view will be confirmed but I’ve attempted to rationalize it away, or my pessimism (like today) will have been alarmism and therefore I get to savor the “victory”.

So emotionally I’ll certainly take that the pattern didn’t persist and maybe the low point is real. BUT I won’t celebrate this “success” by eating anything as I hope to stay on my extreme diet to really have punched through my stuck value before I go on the vacation and start blowing all this.

Speaking of that, of course, I’m already whining I’ll gain 10 pounds during vacation (that’s about a pound/day). Of course that isn’t biologically possible. That would imply consuming about 5500 calories a day and getting zero exercise. Now exercise will be low (driving for 10 hours is hardly going to get my heart rate up) but that level of consumption (even greasy southern food or snacks in the car) is unlikely to be that bad. More likely I’ll do something like 1000 calorie/day excess which would really work out to less than 5lbs gain.

But I do believe my current weight isn’t “real”, that extreme dieting and exercise cause certain unsustainable changes (so-called water weight). This is what people see when they start weight loss, a big immediate improvement, which is merely the body’s rapid adjustment to the change, a one-time only drop. That drop is also easy to reverse. So literally I will probably gain 5lbs in the first two days of the trip, which is really probably just returning to my “true” (aka, sustainable) weight, and then I’ll gain in addition to that.

So when I return and hit the scale for the first time in nearly two weeks, I expect a disastrous gain. BUT, I also expect some of that will be quickly erased and the rest will take some work. And I’m hoping that recent gain will be easy to reverse. So this will be an experiment, to see how well my “prediction” holds, because this kind of event (a break in diet/exercise discipline) is going to happen and it’s how I respond to it that matters. I think this is a big part of how most (almost 100% according to studies) of people who lose weight eventually gain it back. It doesn’t happen all at once or gradually either. It happens in little spurts (the family pigout at a holiday, the special occasion for eating out, the injury or laziness cutting back on exercise for week, etc., a couple of pounds here, a couple of pounds there, and two years later that all adds up and you’re back where you started).

And that is my point of obsessing about these statistics and graphs. I think the idea that for the rest of my life I will have the discipline to avoid short-term gains is totally unrealistic. So what I need instead is the warning system (all this data analysis), the awareness, and the commitment that when I do backslide I immediately compensate for it. You can’t wait two months until you discover your pants are too tight, then think, “oh, need to get back to dieting” while simultaneously getting out your old fat pants. No, you have to compensate right away. OK, had some fun, put on some pounds, get them off now. That I think is the key to weight maintenance after weight loss. Your eating habits are going to shift back up from the diet levels but hopefully you’ll control those better than you did in the past (I really don’t believe it’s realistic to go for what all the nutrition scolds say that you’ll have to change your entire outlook, people don’t do that, at least continuously). No, the change you realistically can make is vigilance (to detect short-term gains) and commitment (to reverse those gains immediately, not in two weeks, but now, starting the day you’re sure the pattern has appeared).

And that’s why all this signal-to-noise and searching for patterns matters.  I need effective techniques to see changes, hidden in a lot of noise, and probably also tied to basic principles (like the big pigout at an event) to conclude it’s time to kick back into loss mode. And a bunch of false alarms (either false positives or false negatives) will be problematic, because the longer a really shift goes uncorrected the harder it will be to correct it.

So hopefully the outcome of this obsession with data gathering and analysis will be that warning system, something I can sustain at low effort, something I can trust as a reliable indicator. So I’ll know when I have to do something.

Then the commitment part comes in a month or so when I finally visit the doc and see whether all this weight loss was worth it. I’m not doing this for vanity or even vague sense of good health, I’m doing it for very specific outcomes, which are statistically only p=0.33 likely to occur. I’m hoping I’m in the 1/3rd where weight loss (actually the easiest solution) does the trick. If I’m not then while there is some benefit to this loss it won’t be my “solution” and I know that will massively undermine my “commitment” to maintenance. IOW, if I have to go the meds route anyway, instead of lifestyle changes being sufficient, then what difference do the lifestyle changes make. Why bother being hungry the rest of my life if it isn’t working?

So there will be a few more months of this thread. It will probably be May before I’m stable at this weight again and then I’ll get my exam and then go from there.

Coincidentally I am now showing a lot of excess skin. Skin needed to wrap a lot of flab doesn’t disappear, at least not quickly. Some people who go through such large changes even get cosmetic surgery to remove that excess skin. This is where I won’t bother. I have enough vanity to actually not like the way I look now (I was fat before, now I look deflated) but not enough to go under the knife (maybe if I were 20 years younger I’d care). The decreased weight will hopefully change my body chemistry and change stress on my joints and if that happens, hurrah, this was a success and I’ll live with some saggy skin.


About dmill96

old fat (but now getting trim and fit) guy, who used to create software in Silicon Valley (almost before it was called that), who used to go backpacking and bicycling and cross-country skiing and now geodashes, drives AWD in Wyoming, takes pictures, and writes long blog posts and does xizquvjyk.
This entry was posted in comment, musing and tagged , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s