Does Diet Coke rot the brain?
Why you should question sensational medical news

[Updated.]

How should we react when the news media report that some food or medicine, previously thought safe, has suddenly been found to be harmful?

An example is the recent study by Pase and others linking artificially sweetened soft drinks (but not sugary drinks) to a sharply increased risk of stroke and dementia. Should we panic?

Well, I am not medically trained, but I do know something about experimental methods and statistics and have been involved in medical research. So let me step through how I react to a news report like that.

(1) Make sure it's not fake news. If the report had come from a glamorous supermodel, or a web site full of "things they don't want you to know," shared by a friend's nephew's uncle's girlfriend's daughter on Facebook, I'd just disregard it. The world is full of medical fake news.

In this case, the research was reported in major news media, they tell us where the research was done and where it was published, and I was able to find it in PubMed immediately.

PubMed is something all educated people should know about. It's a free online index of medical research. It gives you summaries of the articles (or in some cases just bibliography entries), but it often has a full text link to the whole article. It also gives you links to related articles.

Even if you're not qualified to understand the research, you can use PubMed to confirm that it was published in a reputable journal, and to see the results in the researchers' own words, not distorted by journalism or gossip.

(2) Look for confirmation and expert commentary. Before we delve into anything technical, we ask the big question, "Are other researchers getting the same result?" In this case, no; nobody had heard of it before; that's why it's a sensational story.

In fact, along with the published research, in the same issue of Stroke, there's another article commenting on it and saying, in effect, "We don't believe it." Well, not quite that strongly. But they see the results as largely unexplained and needing further study.

Of course, it's often a good idea to ask your own doctor. (My doctors know I do a lot of this!) A practicing physician will have a good feel for whether a new discovery is going to hold up or not.

Now let's get more technical...

(3) Remember that "significant" doesn't mean "proven." There is random variation in all experiments — no two groups of people are exactly alike — and sometimes the random variation is big enough to "throw" the results of a study. Oversimplifying just a bit, the usual statistical techniques leave about a 5% chance of this happening. So there's about a 5% chance that any statistically significant result still isn't real. (Often, the researchers publish p values that enable you to tell whether the risk is near 5% or appreciably less. And we can fight the frequentist-Bayesian battle somewhere else; I know I'm making a long story very short.)

(4) Think about whether the experiment was biased. This particular study was controlled for obesity and a lot of other things — otherwise the result would have been explained away by the simple fact that only overweight people drink diet sodas.

A colleague reminds me that this controlling probably wasn't perfect. The experimenters matched the two groups for everything they could — but some important difference may still have been overlooked, or the information might not have been available.

Another source of possible bias, common in medical studies, is simply that one group gets more and better medical care than the other. That's probably why we get scattered reports that one thing or another is slightly associated with Alzheimer's disease, and then nobody confirms it. Some people are more likely to be diagnosed than others, simply because they get more medical care, and those are also more likely to receive unrelated medical treatments for other things.

(5) Look for a dose-response effect. If X causes Y, then more X should cause more Y. If it doesn't, you're probably looking at random variation.

And that's where, to my eyes, this study breaks down. Looking at the full text (this link may not work for everyone), in Table 2, the dose-response effect is either weak or not there at all. The researchers compared people who drank diet sodas occasionally (0 to 6 per week — presumably 0 doesn't quite mean 0) with those who drank more than 1 per day. The risk of stroke was only a few percent higher for the second group, and in one case it seemed to be lower. With dementia (Table 3), the dose-response relation was more convincing, but the statistical significance went away when the patients were matched for a number of common medical conditions (that might lead a person to go on a diet).

So what am I going to do? Mostly, keep my eyes open for other studies that might confirm or explain these results. If two or three more studies back it up, I'll stop drinking diet soft drinks.

Just a couple more thoughts. Although research results are unclear, we already have grounds to suspect that diet soft drinks might be fattening even though calorie-free. How do they do that? Apparently, the sweet taste signals the body to store calories and stimulates the appetite. This is not solidly confirmed, but it is plausible, and it could introduce confusion into a study like this.

Also, during the time period of the study, as far as I can tell, virtually all diet soft drinks were sweetened with aspartame, about which there is slight but lingering uncertainty. (It gives some people migraines, which is, I suppose, a bad sign.) Today's diet soft drinks often use sucralose (especially Diet Pepsi, the one I drink the most). So we're not looking at the same situation today.

Bottom line? Wait for more results. But of course if you don't like diet soft drinks, you don't have to drink them.

Permanent link to this entry

"Idiot lights"

Fifty years ago, people were lamenting how the gauges on car dashboards were being replaced by "idiot lights" that would light up without really telling you what was wrong. (These were lights like "Hot" and "Oil." The universal, incomprehensible "Check Engine" light came later.)

Today we have a new plague of the same thing, thanks to microprocessors in appliances.

Our Keurig coffee maker demanded de-scaling (cleaning) the other day, so I bought Keurig's "only recommended" cleaning solution (a $16 bottle of citric acid solution that must have cost 10 cents to make) and ran the cleaning process. The "DE-SCALE" message keeps popping back up. I've done some further cleaning by hand, and today I'm running vinegar through it.

If it doesn't stop making these demands, I'll have to ignore them.

Fundamentally, the Keurig is violating a principle of communication called Grice's Maxim of Quantity: "give the right amount of information." The trouble with "idiot lights" is that they don't tell you enough to enable you to fix the problem. This one is as idiotic as any.

Permanent link to this entry

Farewell, old tree

I'm ashamed to say I never paid much attention to the enormous tree just southwest of the Boyd Graduate Studies Research Center — the building in which I worked for nearly 30 years — until I passed by last week and saw that it had been cut down. Presumably, the storms damaged it.

I think it was a white oak. What I'm sure of is that the stump was four feet in diameter.

This tree must have been there before Boyd was built in 1968. Before the building was built, the site was an amphitheater (which explains why the building seems to be sitting in a hole) and this tree must have adorned one of its four corners. Tempora mutantur...

Permanent link to this entry

Department of Linguistics

Yesterday the University Council unanimously approved the creation of a Department of Linguistics at the University of Georgia. All that remains is for the president of the university to sign off on it.

UGA has been giving degrees in linguistics since the 1970s (I was their second B.A.). All this time, linguistics has been operating as an interdisciplinary program between the various language departments. In recent years it has had more students than, for example, the Department of History.

But linguistics is not interdisciplinary. It is a well-defined subject, not just the intersection of others. Linguistics is the scientific study of human language. It is not just the historical or dialectological study of one or another particular language.

Interdisciplinary programs are legitimate. Artificial Intelligence is an interdisciplinary program because everything we do is also either computer science or psychology or philosophy or some combination of the three. African-American Studies is interdisciplinary because everything they do is also history or sociology or something. But linguistics is linguistics; it is not the overlap of the language departments; it is not any more interdisciplinary than economics, for instance. It should have been a department all along.

Particularly poignant for me is the fact that I had my whole career here, from inception to retirement, without ever having "Linguistics" in any of my academic titles, because there was no such department.

Permanent link to this entry

Spiral galaxy

One of the few galaxies whose spiral structure is visible in a telescope is M51, which you see here. I've photographed it before, but this is probably my best image of it. Celestron 8 EdgeHD with f/7 reducer, Nikon D5300, stack of six 3-minute exposures autoguided on a CGEM mount. The processing was with DeepSkyStacker and PixInsight, with increased color saturation in the highlights (distinguishing yellowish dust from bluish stars) and a bit of deconvolution to make the star images rounder.

Permanent link to this entry

Two spirals having a fight of some sort

"Two spirals having a fight of some sort." Imagine those words spoken in a baritone voice with a Dutch accent, and you will know how the late great Bart J. Bok, expert on galaxy structure, described these galaxies in a lecture at the International Science School for High School Students in 1973. (I was there.)

These galaxies (NGC 4038 and 4039) have in fact suffered a collision. Imagine, however, not cars colliding, but loose swarms of bees colliding; for the most part, the individual stars did not collide, though their movements were affected, as was the density of interstellar gas. That triggered regions of active star formation, which you see as bright beads here.

These galaxies are sometimes called the Antennae because there are two faint streamers coming out of them, barely visible in the picture, like the antennae of an insect.

Same technique as the picture above, but a stack of 10 3-minute exposures.

Permanent link to this entry

Interoperability of TortoiseSVN and TortoiseGit

Subversion (SVN) and Git are two source-code control systems for storing changing versions of computer programs as you develop them, with the ability to retrieve past versions and compare them for changes. Of the two, Subversion runs on your own server (I recommend CollabNet's free server software) but Git usually relies on GitHub, a cloud storage site.

What if you want to use both?

I was using Subversion together with TortoiseSVN, which is a Windows Explorer extension that makes it easy to manage files by right-clicking on them. TortoiseSVN even has icon overlays that put red or green marks on folder icons to tell you whether they have uncommitted changes.

To my delight, there's also TortoiseGit, which shares some code with TortoiseSVN and uses the same icon overlays (so if you've sorted out problems from having too many of them, the problems won't reappear).

Here are some notes on how I got them to work together well. The key problem is keeping each of them from putting much on the right-click menu of folders that belong to the other one. I don't want to do Git operations on SVN folders or vice versa.

(1) Get TortoiseSVN working the way you want it.

(2) Download and install the Git client software (Git proper, not TortoiseGit). Take the defaults. Uncheck "Windows Explorer Integration." If you previously installed it with that item checked, uninstall and reinstall.

By the way, Git gives you some handy things — not only command-line utilities, but also a UNIX-like bash shell for Windows (courtesy of MinGW) that I haven't explored, in which there are a lot of GNU utilities. There's a GUI for Git, which we'll ignore.

(3) Spend half an hour reading about how Git works. It's different from SVN. The biggest difference is that committing a change is a 2-step process: "commit" means stage the file for sending, and "push" means send it to the server. "Pull" means download changes and apply them, two steps that can be separated ("fetch" and "merge").

In Git, a "repository" is any folder that is managed by Git (and has a .git subfolder); in SVN, a "repository" is what resides on the server. Accordingly, the equivalent of an SVN Checkout operation is to "clone" an existing Git repository from the server to your PC.

(4) Download and install TortoiseGit.

(5) The unobvious part: Make settings in TortoiseSVN and TortoiseGit. You can get to each of them by right-clicking anywhere and choosing "TortoiseSVN" or "TortoiseGit" respectively, then "Settings."

What you want to do is limit what is in the topmost right-click menu, so that (apart from "TortoiseSVN" and "TortoiseGit") you will only see operations that are valid on the current folder (which may be an SVN folder, a Git folder, or neither).

In TortoiseSVN, choose Settings, Context Menu, and make sure only "Commit", "Update", and "Add" are checked. (They only show up when you right-click on or inside a TortoiseSVN folder.) All the other menu options are accessed under TortoiseSVN.

In TortoiseGit, choose Settings, Context Menu, and make sure only "Push," "Commit," and "Pull" are checked. (Again, they only show up when you right-click on a TortoiseGit folder, and all the other menu options are still there, under TortoiseGit.)

(6) Reboot.

Important note: In Subversion, a red icon overlay means there are changes that have not been sent to the server. In Git, a red icon overlay means there are changes that have not been committed, and it turns green when you commit them, whether or not you have pushed them to the server. Remember to push!

Permanent link to this entry

Visual Studio 2017 hangs at "Build started..."

Symptom:

In Visual Studio, you start to compile and run a program in debug mode, and Visual Studio hangs, unresponsive, with "Build started..." showing at the lower left corner.

Workaround:

Debug, Options, Debugging, General, uncheck "Enable diagnostic tools while debugging." You will no longer see memory and CPU usage graphs during execution.

Notes:

(1) Similar problems have been reported with earlier versions of Visual Studio. Myself, I have only experienced it on one computer (an Azure virtual machine), not others.

(2) Updates to Visual Studio 2017 are frequent. Check Tools, Extensions and Updates..., (left column) Updates, Product Updates.

Permanent link to this entry

Curation of Internet advertising

TV stations and newspapers control their ads; some human being knows what is being advertised and can block ads that are unusually obnoxious or unsuitable for the context. Advertising on the Internet doesn't work that way. The ads on a web site are usually put there by a third party, unconnected with the web site owner, who pays for the privilege of putting anything there, even malware.

That may be about to change. An ad blocker may be about to be built into Google Chrome, currently the most popular web browsing software. It won't block all ads, only those that are violate certain standards, such as ads that make noise or pop up in front of what you are reading.

I'm sure one of their motives is to block the ones that claim to be Chrome updates or error messages!

Permanent link to this entry

What they forgot when they legalized cannabis

I'm against legalization of cannabis; I think it's not as safe as most people think. But given that some localities are legalizing it, I see something big that's missing: standardization of strength.

Alcoholic beverages are required to be labeled with their alcohol content. Why not label cannabis products with their THC concentration?

I know this would make manufacturing a lot more expensive, but it would also reduce a hazard. Cannabis overdoses aren't fatal, but they are certainly unwelcome and can be costly, and there's a steady stream of reports of them out of Colorado.

Admittedly this leaves open the issue of other cannabinoids. As I said, I don't think the stuff is safe enough for recreational use. Medicinal use is a different thing — but not as a smokescreen for recreational use. If it's medicine it ought to be marketed like medicine, not as a recreational product with a nudge-nudge-wink-wink.

And yes, I did notice the date of today's entry.

[Update:] Although I don't have the details, I'm told some kind of strength labeling has finally been adopted in Colorado. But I also heard a Coloradan pointing out that there is a serious problem with secondhand marijuana smoke — people don't want to be drugged against their will! I still think the legalization of cannabis was poorly thought out and poorly implemented.

Permanent link to this entry

The life of a linguist

I just had a funny experience in an electronics forum. Someone posted a picture of an unidentified piece of equipment and said, "All that I can find about this on the Internet is in German."

So I said, "Point me to the German page and I'll read it."

So he did. It's a picture of the same equipment, and under it, the words, in German, "I am looking for someone who can tell me what this piece of equipment is."

Permanent link to this entry

Forgotten, and deservedly so

Sharon showed me some of the BBC "Father Brown" episodes that are set in the 1950s, when Fr. Brown is older and has several regular sidekicks.

I'm impressed by the smoke-free or nearly smoke-free pubs. That's an obvious anachronism, but it must have made the series easier to film. A real pub in the 1950s or even the 1970s was a very smoky place.

Then it struck me: Many of today's adults, including smokers, have never experienced a really foul-smelling smoke-filled room such as were common up to 1975 or so — a room in which people have been smoking constantly all day for years, without adequate ventilation, so the dominant smell is not burning tobacco but something worse, almost sewer-like.

Not only pubs and bars but also offices used to be like that. My mother worked in such an office in the 1960s. The last one I encountered was at Connecticut Bank and Trust around 1980, and I changed bank branches and wrote them a letter telling them why.

Those are a thing of the past, not only because of indoor smoking bans in the 1990s (spurred by convincing evidence that secondhand smoke is hazardous), but also because of improvements in air conditioning that took place a good bit earlier.

And now they're forgotten, just as the stench of Victorian city life has been forgotten. Most of us, when reading Sherlock Holmes, do not think about what Baker Street smelled like. And that is a good thing.

Permanent link to this entry

Jupiter, Europa, and Europa's shadow

On Easter evening (April 16), Jupiter's satellite Europa passed in front of the planet and cast its shadow on it. Here's the best of several pictures I took. 8-inch telescope, 3x extender, Canon 60Da in movie crop mode, same as for several of the Jupiter pictures below.

Permanent link to this entry

Permanent link to this entry

Jupiter again

Another image of Jupiter with the 8-inch, same equipment and technique as the one below (scroll down). This is the kind of quality I can get reliably when Jupiter is as close to the earth as it is now.

You can see the Great Red Spot peeping around the edge at the lower left.

Permanent link to this entry

Lenovo, le vieux: Back to the old

Although others have ordered the same make and model and gotten good results, the Lenovo Z50 that I bought from Microsoft has had to go back for a refund. It has several symptoms of unreliable hardware, including a touchpad and Wi-Fi adapter that behave erratically. It may have gotten banged up in transit. I've gone back to my 2011 Lenovo Z570, which — with newly replaced keyboard, touchpad, and hard disk — is holding up fine.

Permanent link to this entry

Jupiter with the 8-inch

Here's Jupiter again, same technique as the one below, but with my Celestron 8 EdgeHD telescope and steadier air. Best 33% of about 4100 video frames.

Permanent link to this entry

Jupiter with the 5-inch

My vintage-1980 Celestron 5 (which had a major optical reworking in the early 1990s and now has a coated corrector plate) continues to serve me well. When the air is unsteady, it can give a sharper image than larger telescope. So it is that on the evening of April 9, around midnight EDT, I took the picture of Io, Jupiter, and Europa that you see here.

I used a Meade ×3 telecentric converter ahead of my Canon 60Da, which functioned as a video camera in "movie crop" mode, which means it used the full resolution of a small area in the middle of the sensor. (Most video modes — all on newer DSLRs — downsample their images to lower resolution because multi-megapixel video files would be too big and too slow to handle.) The picture is a stack of the best 50% of about 3600 video frames, aligned with AutoStakkert and sharpened with RegiStax 6.

Permanent link to this entry

Down south on the moon

Last night I also got a good view of the southern region of the Moon, south and (selenographic) west of the crater Tycho. These were taken with the same equipment, leaving out the ×3 converter, so the image scale is one third as big.

Here you see the region from the dramatic crater Bailly (on the right) to the lava-filled crater Wargentin (the one that doesn't seem to have been hollowed out like the others) at the far left:

And here you see a wider field whose most conspicuous feature is the large crater Schickard:

Permanent link to this entry

Who's right, Tolkien or Star Wars?

Today I want to give my non-Christian readers an important taste of what it is like to be a Christian. The following is reprinted, with his permission, from a paper recently presented by my good friend and mentor Donald Williams.

"The thought pierced [Sam] that in the end the Shadow was only a small and passing thing; there was light and high beauty forever beyond its reach."

The Lord of the Rings is not just a story of Good versus Evil. It is a story in which good and evil are opposed, but they are not equal. They are not just the light and dark "sides" of the same Force, as in the Star Wars universe. The Shadow which oppresses us and seems invincible is in the larger scheme only a small and passing thing. Final victory is forever beyond its reach, just as the clouds billowing from Mount Doom could never rise above the atmosphere to really put out a star in the heavens. The star and its light will remain when the wind has blown all of Sauron's gloom away. The star wins! Symbolically, Good wins. No wonder Sam feels encouraged.

How can this be so? In a naturalistic universe such a claim would be meaningless because good and evil are only subjective human perspectives and no finite human perspective can ever be the final word about anything. In a universe that was "spiritual" but not theistic, i.e., a pantheistic world like that of Star Wars, the claim would be false. Neither Good nor Evil can be ultimate there because both are merely aspects of the One reality that is all-encompassing. There can be no final victory for either, ever, because only a Sith deals in absolutes.

Only in a theistic universe, a universe created by a God very like the God of the Bible, can the thought that comes to Sam be meaningful and true. In that world, goodness is not arbitrary but is rooted in the character of the creator God. Evil arises through the abuse of the free will God has granted to his higher creatures and is tolerated for a while as the price of their significance. But it is derivative, not original, and is thus doomed to eventual defeat because God as the creator is the greatest Power that is or can be. Thus the ultimate triumph of Good is guaranteed by the very nature of the world as created and its relationship to its creator.

Permanent link to this entry

Why should integers have all the fun?

Permanent link to this entry