Good science/bad science

On a private email list someone brought up the subject of "good science" vs. "bad science", and suggested a couple of other categories: "ugly science" and "not even wrong science". (The latter refers to a remark reportedly made by the acerbic physicist Wolfgang Pauli when asked about a certain piece of research he evidently considered utterly ridiculous.)

I'm not especially comfortable with such categories. Taking the categories of faulty science first, a little thought suggests that there are a number of ways that some scientific idea or hypotheses can be wrong. A hypothesis can simply be mistaken, though plausible given available knowledge at some point in time. Even the very best scientists can make such mistakes, especially regarding phenomena that have not been amenable to proper research at the time. The concept of a "luminiferous aether", prior to the Michelson-Morley experiment, is a suitable example. Even Einstein made this kind of mistake, for instance in believing that the universe was static rather than expanding, before Edwin Hubble demonstrated otherwise.

But "bad science" is something worse than mistaken. It is an idea or hypothesis that is based on errors in methodology or reasoning or understanding that a competent scientist simply should not make, given available knowledge at the time. "Intelligent design" may be the canonical example of this sort of thing, but there is no lack of other examples throughout the history of science.

"Not even wrong science" is even worse. It's usually the product of an earnest, well-meaning individual who has an enthusiasm for some branch of science, but hardly any actual training or understanding of the subject. Examples of this sort of thing include cosmological theories involving "dynamic energy vortices", "proofs" that the theory of special relativity is obviously wrong, etc. Alternatively, some would place in this category any purportedly scientific theory that has no testable predictions -- superstring theory being a favorite (alleged) example. I won't get into that debate right now.

But then where do we put other kinds of "bad science"? For instance, plausible results which are nevertheless fraudulent because the experimental data has been fudged or simply fabricated. Some, but not all, of Korean stem cell scientist Hwang Woo-suk's work is the currently best-known example. Or how about the "results" of scientists employed by tobacco companties, which failed to find a link between smoking and lung cancer?

Many people use the term "junk science". Where does that fit in? Unfortunately, it's used rather loosely, sometimes in good faith and sometimes not, by anyone who wants to disparage unwelcome scientific claims. Both believers and skeptics of human-caused climate change use the term to describe the claims of their adversaries.

But enough about "bad science". You can find many lists of alleged examples, for instance at the site appropriately called Not Even Wrong. (That's the work of Will Kinney, rather than the perhaps better-known blog dedicated to arguments against superstring theory.) See also the newwpaper column called Bad Science, by Ben Goldacre.

But I've gone on much longer than I intended to regarding bad science. What about "good science"? Can we consider it to be any science that is fairly generally accepted as being correct? I'm not comfortable with that either. The problem is that there's no good set of operational criteria for certifying some scientific theory to be "generally accepted as correct". Indeed, for any given scientific theory or claim, there's a whole spectrum of confidence about whether the results can be considered as "proven".

Of course, there are certain necessary conditions a theory must meet in order to be considered "good science". It needs to be internally consistent and consistent with other "facts" that are also considered to be "known". (That can be a can of worms.) It needs to be falsifiable (i. e. there has to be some conceivable experiment that could rule out the theory). There ought to be at least some evidence that actually supports the theory. And so forth.

Finding sufficient conditions, however, for a theory to be considered "good science" is a lot harder. I'll present a list, in a moment, that offers several examples. The overall point is that for any given theory, it is often reasonable to regard both the theory and its negation as plausibly correct to some extent. Which means that both the supporters and opponents of a theory can be reasonably regarded as advocating "good science", even though at least one side is actually wrong... if and when we could determine the "real" truth.

And so, I tend to regard as "good science" any reasonable, plausible hypotheses that meet the necessary conditions and have "some" evidence in their favor, even if they haven't yet been fully "proven". Do we need to have a name for this category, such as "good but not fully proven science"? I don't know, because in fact most hypotheses which are still being researched naturally reside in this category -- even if they are pretty generally accepted as correct.

Herewith, some examples (I won't clutter this up with links to Wikipedia, but you may consult it, or the reference of your choice, for names and terms you're not clear about):

1. Black holes. At first almost everyone, even Einstein, thought this was a crazy idea. Eddington nearly destroyed Chandrasekhar's career over this issue, though the evidence for black holes, lately, has been pretty darn good. But there are still doubters, and very recently a proposal has been made that could give an alternative account of black holes, and the controversial theories of dark matter and dark energy as well. See here.

2. Dark energy/cosmological constant. Although a small cosmological constant used in the Friedmann equation gives a very good fit with universe expansion data from supernovae, other (still quite controversial) observations of very distant gamma-ray bursts do not fit. And there are alternative accounts -- see above, and for a different one see here.

Further, there is no decent theoretical explanation of a small cosmological constant.

3. Dark matter. The evidence for this is very good, and of many kinds. Yet many people keep trying to come up with alternatives. Again see above. A very reputable physicist, Jacob Beckenstein, has recently claimed to have reconciled MOND (modified newtonian dynamics) with relativity so as to provide another viable alternative. There's an article on this in NewScientist for 29 April, though I find that magazine has a penchant for pushing iconoclastic theories. See also here.

4. Inflationary cosmology and hypotheses about its antecedents. There are sound mathematical theories for such things, but little (for inflation) or nothing (for antecedents) in the way of physical evidence. The "standard" hypotheses is the Big Bang, but there are many variants and alternatives, associated with big names like Hawking, Steinhardt, Linde, Turok, etc. Unlike the previous examples, there's scant evidence, in spite of much elegant theory. Such theories need a category other than good/bad/wrong/ugly/not-even-wrong.

5. Quantum mechanics and determinism. Some quite reputable physicists keep trying to find some sort of determinism underlying QM. Most recently nobelist Gerard 't Hooft (NewScientist 5 May and here). Of course, this presents issues with the Conway/Kochen free will theorem. See here and here.

6. The Alvarez asteroid theory to explain the Cretaceous-Tertiary mass extinction. Since the discovery of the Chicxulub impact crater, people have become pretty convinced of this theory, despite heavy early skepticism. Yet evidence keeps turning up that the impact crater was formed long (ca. 300,000 years) before the extinctions began. See here.

There are many more examples, of course, including life sciences and medicine -- contentious stuff like megadose vitamin C, (denial of a relation between) AIDS & HIV, RNA-world, panspermia, etc. And that's without even getting into the squishy "sciences" like psychology, sociology, economics.

Having categories like good/bad/ugly may sometimes be helpful -- but they're definitely over-simplifications. Reality is messy.

Reversing the cell division cycle

Most eukaryotic cells are "diploid", which means that each cell contains two copies of most of the chromosomes in the total genome (excepting the sex chromosomes, which in mammals are known as X and Y). The two copies of each chromosome are not entirely identical, though each contains the same genes. The difference is due to isolated variations that may occur within genes or outside genes in the non-coding "junk" DNA of a chromosome.

Humans, for example, have 22 paired chromosomes (besides the sex chromosomes). Both chromosomes in each pair have the same genes, but slight differences exist since in each pair is an almost exact copy of one chromosome from each of the parents. (The copies are not quite exact, due to mutations that may have occurred since the paternal sperm fertilized the maternal egg.)

It is important that every daughter cell that results from cell division continues to have both variants of each chromosome. This guarantees that every cell in the body continues to have the same genome, in which every gene is represented by a copy from each parent. For example, if 1 and 1* designate the two copies of chromosome number 1, then both daughter cells should have a 1 and a 1*, rather than two 1s in one daughter and two 1*s in the other.

Of course, this only applies to organisms (usually multicellular) which undergo sexual reproduction involving the fusion of an egg and a sperm. Cells of such an organism are "eukaryotic", which means that they have a nucleus and a genome consisting of several chromosomes which have a fairly elaborate internal structure.

It should not come as a surprise that the process of cell division required in order to keep all the genetic information straight is fairly complex. The overall cell division process is referred to as the "cell cycle". Sometimes the whole process is referred to loosely as "mitosis", but cell biologists normally reserve that term for the final phase of the cycle, in which all parts of the dividing cell are apportioned between each of the new daughter cells.

There are three other phases of the cycle before mitosis. Omitting most of the details, all that needs to be said is that it is in the second phase when the DNA making up the chromosomes is actually duplicated. This is a fascinating process in itself, but not relevant here. All you need to know is that after the DNA has been duplicated, each and every chromosome consists of two identical copies (unless an undetected copying error occurred). These duplicate copies are joined together at a point in the middle called a "centromere", and each copy in the pair is called a "chromatid". So at this point, there are actually four copies of each chromosome type (other than sex chromosomes) -- two copies from each of the organism's parents.

Later, in the mitosis phase, each pair of chromatids will be separated in such a way that each daughter cell inherits half of every pair of (identical) chromatids. All the complexity of the process lies in the mechanics of how this is accomplished. Basically, what happens is that all the chromatid pairs migrate to a plane roughly in the middle of the cell. At the same time, two spots known as "centrosomes" form on the cell membrane on opposite sides of the central plane. From each centrosome a network of threadlike "microtubules" is constructed. Eventually, every chromatid is attached to just one of the two centrosomes by a sort of loosely woven rope of microtubules. Once all of this is in place, the connection is severed between chromatids in each pair, and every chromatid begins to "climb" up its connection towards the centrosome it's attached to. Finally, the central part of the cell contracts, new nuclear envelopes form around the chromatids in each half, and at the end the two halves separate. And then the cycle begins again in its initial phase.

Given all this complexity -- most of whose details have been omitted -- I found it quite interesting, and surprising, that new research indicates this whole process can be reversed! This doesn't seem to happen naturally, as far as we know, but it can be made to happen by manipulating some of the proteins (known as "cyclins" and "kinases") that regulate the cell cycle.

Here's the press release that announces the pubication in Nature of the research: Rewind, please: OMRF scientist reverses process of cell division. Unfortunately, the press release itself is really lame. It explains almost nothing about the actual research -- so if one is at all curious about what is actually meant by "reversing" the cell division process, one is quite disappointed.

The research article itself, whose lead author is Gary Gorbsky of the Oklahoma Medical Research Foundation, can be found here. Unfortunately, unless you have a subscription to Nature (or are willing to pay $30 for the privilege of having one week's access to the article), you won't be allowed to see it. But at least you can view the abstract here.

Even that won't be too helpful unless you're really up on your cell biology. So I'll try to explain in something a bit closer to English what the research is actually about.

As noted above, the transition between phases of the cell cycle is regulated by proteins called "cyclins" and "kinases". Within each phase, the appropriate genes get switched on to bring about whatever activity is to occur in that phase. In particular, the mitosis phase is kicked off by the appearance of a "cyclin-dependent kinase" called Cdk1 along with a cyclin molecule called, simply, cyclin B. Cdk1 is an enzyme that catalyzes the attachment of phosphate groups to other molecules. But it's active only when it has a cyclin attached, in this case cyclin B.

The cyclin molecules are not as robust as the kinases, which is a good thing, because it is their eventual breakdown, after all the work is done, that deactivates the kinase and allows a new phase of the cycle to commence.

What the research found is this: After the mitosis phase has ended ("mitotic exit"), but before the two new cells actually separate, the process can be returned to the middle of the mitotic phase by manipulating Cdk1 and cyclin B. What you do is block the degradation of cyclin B but use an inhibitor to disable Cdk1. Mitotic exit still occurs. But if you quickly remove the Cdk1 inhibitor, the process reverses. The membrane between the new cells reopens, the new nuclear envelopes break down, the microtubules reappear, and the chromosomes move back to the middle of the cell. Got that? The bottom line is that the degradation process of cyclin B provides directionality for the later part of the mitosis phase.

Well, all in all, it's a really fascinating bit of bio-hacking. Is this somehow, you know, useful in any way? Well, at least it's very interesting new knowledge about how fundamental cellular processes work. If we learn how to control the processes, we may figure out how to intervene in disease conditions when cells malfunction -- as in cancer, for instance. After all, cancer involves runaway cell division. If you can reverse the whole division process -- selectively, where it should not be happening -- you may have just found a cure for cancer.

------------------------
See also:
Gary Gorbsky home page
Gorsky Lab Home Page
Cell division rewinds
------------------------

Tags: biology, mitosis, cell cycle