Open access series: Total open access.

Last week (over hyar), I presented and critiqued three arguments in favour of what I (over hyar) called "the status quo". All three were negative arguments, and I suggested that this was a weaker strategy than arguing positively for the benefits of the status quo. The first argument I called the "cost argument", and it asserts that alternatives to the status quo will be more costly than maintaining the status quo. The problem with this argument is that it's not clear why we should accept that it's bad to pay the price/lose the benefits that we would in changing from the status quo. The second argument I called the "academic argument", and it asserts that alternatives to the status quo are unable to play an important academic function. I argued that neither of two important functions -- peer review and the "elitist function" -- are necessarily best served by the status quo, nor are necessarily incompatible with an alternative to the status quo. Finally, the third argument I called the "relevance argument", and it asserts that, as there's no need to change the status quo, it should not be changed. I argued that this is both question-begging -- in that the argument just asserts that there's no need to change the status quo -- and contains a highly questionable premise -- as there is no demonstration of the needed connection between need and obligation.

This week, I'm considering arguments in favour of TOA -- total open access. Here's what total open access is, as I've defined it:

TOA would have no or exceedingly minimal charges for access and extremely high ease of access. In order to read most published research, one would go to a freely-accessible database that would contain all relevant research publications. This database would be easy to use and complete; and would also allow one to directly access the materials found, in either electronic or dead-tree form. All researchers would be required to have their publications available through this database.

To have the relevant contrast in hand, here's how I defined "improved access" or IA:

IA would have modest charges for access and high ease of access. That is, the actual process of searching for published research would be highly similar to that of TOA, with one exception, namely that no one would be required to have their publications available through this database. To incentivize the process, researchers would be rewarded for participating, but not punished for failing to. After all, at the end of the day, the published work is their own and, if copyright means anything, it means that one can decide how to release what one has created. (Or even not to do so at all.) There would be more significant access fees than under TOA, but far less than under the status quo; and this would follow simply from the fact that, instead of subscribing to multiple databases and publishers in order to access all relevant research, individuals and institutions would only have to subscribe to one.

Arguments in favour of TOA seem to fall into one of two broad categories, each of which subdivides into two further categories. Here they are, schematically:

1. Intrinsic Justifications
   1. Entitlement
   2. Empowerment
2. Instrumental Justifications
   1. Benefits
   2. Costs

Illuminating, I know. Here's what this all means.

(A) is the category of arguments that try to show there's something that is valuable in itself gained from TOA. That is, regardless of anything else, gaining this whatever-it-is is a sufficient gain to justify changing to TOA. (A)(i) identifies this whatever-it-is as something like respect for what we deserve or what we are entitled to. That is, according to (A)(i), TOA respects that we are entitled to research or knowledge or free access to knowledge or some such. (A)(ii), by contrast, identifies the whatever-it-is as some sort of empowerment. That is, according to (A)(ii), TOA empowers citizens by giving them a new and important sort of ability. If you like, (A)(i) can be read as defending TOA on the grounds that it respects a certain kind of negative right (not to be interfered with regarding my entitlements), while (A)(ii) can be read as defending TOA on the grounds that it grants a certain kind of positive right (to be a better sort of person, more knowledgeable, better at important judgements, etc.).

(B) is the category of arguments that try to show there's something valuable gained from TOA, but not from TOA itself. Instead, TOA gives us something that is valuable for other ends we may have. (B)(i) identifies this something as a broad class of benefits, usually economic, but also things like increased access for readers, which, for researchers, could lead to increased numbers of citations and thus increasingly fruitful debate, and, for professionals, could lead to increasingly better ability to perform their professional obligations (think of physicians having greater access to cutting-edge medical research). (B)(ii), by contrast, identifies this something the lack of a broad class of costs, again usually economic, including things like subscription payments and secure servers to block unauthorized access to proprietory data, and so on. So, (B)(i) defends TOA on the grounds that it allows us to gain, and (B)(ii) defends TOA on the grounds that it prevents us from being harmed.

These are not strict categorizations, of course. This is just a way for me to organize a response to the defense of TOA. So, addressing these in turn.

(A)(i), as presented thus far, has a pretty significant lacuna. It has to be shown that we are entitled to what we get under TOA. This is clearest if I look at a particular example of the argument.

1. Taxpayers are entitled to access to university research.
2. Therefore, taxpayers should have access to university research.

Obviously, this argument has a giant hole in it: it doesn't explain why taxpayers have this entitlement. It's a pure non sequitur. So, something like the following has to be given instead:

1. Taxpayers partially fund university research.
2. Taxpayers are entitled to access to what they partially fund.
3. Therefore, taxpayers should have total open access to university research.

The problem with this version, though, is that there's still another lacuna, in the move from (1) to (2). How do we get from partial funding to total entitlement? Surely taxpayers are just entitled to the part that they fund. Of course, one could argue that taxpayers' contributions are not, in principle, separable from the contributions of other sources. And this is probably true, but it still doesn't solve the problem. For, while we could go from the claim that taxpayers' contributions are not separable to the claim that taxpayers are fully entitled to the research, we could just as easily go from the former to the claim that taxpayers are not at all entitled to the research. There's no given reason for moving from partial contribution to full entitlement over partial contribution to no entitlement.

There's another problem with this sample of (A)(i)-style argument, namely why contribution converts into entitlement at all. It certainly doesn't in many other venues, particularly when the contributions can be construed as gifts. If, however, the contributions are construed as investments, then there is something to be said for one being entitled, but usually the invested-in party has some choice as to whether to accept the investment or not, and thus the resultant entitlement claim or not. It's certainly not clear that researchers always have that option.

But, it could be objected that these are quibbles, and quibbles with a particular version of the (A)(i) argument. This is probably true. So, let's look at the argument generally:

1. A's have φ'd research.
2. A's are entitled to access to what they have φ'd.
3. Therefore, A's should have total open access to research.

This will cover both the particular argument given above, and any other argument that could fall into (A)(i)'s scope. In the spirit of fairness, I'll allow that (1) and (2) are true, for some value of A and of φ. Even if we grant that, though, we have a clear problem: (3) doesn't follow. (3) substitutes "total open access" for "access" -- what strictly follows would be this:

3. Therefore, A's should have access to research.

In order to repair this logical gap, (2) must be amended as follows:

2. A's are entitled to total open access to what they have φ'd.

But that amendment of (2) begs the question against IA. It assumes that φ-ing grants A's title to total open access instead of some level of access. If TOA were somehow the "default" position, then this would be a legitimate presumption; but, without argument, it cannot be assumed that TOA is this default. (For convenience, I'll call this the "totality presumption", or TP.)

So, (A)(i)-style arguments only work if they presume pro-IA arguments are failures. This may turn out to be a legitimate presumption, but it has yet to be shown. Moving on, then, to (A)(ii)-style arguments (the discussion of which should not be as lengthy, as I can piggy-back on the above discussion).

(A)(ii)-style arguments are going to broadly be similar to (A)(i)-style arguments. They will look something like this:

1. If A's had access to research, then A's would be (able to) φ.
2. A's should be (able to) φ.
3. Therefore, A's should have access to research.

This argument presents three problems, just with the logic. First, it has the same access vs. total open access issue. If (1) and (3) are amended to include "total open access", then this is question-begging against IA -- once again, good ol' TP. Second, though, the argument as given is fallacious: it affirms the consequent.

According to standard sentential logics, there are two legitimate patterns of inference that start with a conditional (an if-then statement), and two fallacious patterns. The two legitimate ones are, first, modus ponens:

1. If A, then B.
2. A.
3. Therefore, B.

And, second, modus tollens:

1. If A, then B.
2. Not-B.
3. Therefore, not-A.

The fallacious ones are, first, denying the antecedent:

1. If A, then B.
2. Not-A.
3. Therefore, not-B.

And, second, affirming the consequent:

1. If A, then B.
2. B.
3. Therefore, A.

Obviously, the (A)(ii)-style argument given fits the latter of these four patterns; therefore, it is fallacious.

How could it be reconstructed to avoid the fallacy? As given, I've built the (A)(ii) argument scheme to treat access to research as a sufficient condition for the valuable ability, φ. I could build it, instead, with access to research as a necessary condition:

1. If A's were (able to) φ, then A's would have access to research.
2. A's should be (able to) φ.
3. Therefore, A's should have access to research.

Or, to treat access to research as both necessary and sufficient:

1. A's had access to research if and only if A's would be (able to) φ.
2. A's should be (able to) φ.
3. Therefore, A's should have access to research.

Either amendment would solve the logical problem, but at a cost. In the necessary-and-sufficient version, we have a second point at which a question is begged, only this time against the status quo. (1) in this version just assumes what the status quo tries to deny, that increased access and some important ability, φ-ing, go hand-in-hand. This may be true, but it needs to be shown, not assumed. (I make no judgement on whether it could be shown; that really depends on the particular φ.) In the just-necessary version, we beg exactly the same question.

So, in its most modest form, the (A)(ii)-style argument affirms the consequent; in its more ambitious forms, it begs the question against the status quo. And there is yet a third logical problem with the argument, which is the move from a descriptive conditional or biconditional (in all versions of (1)) to a normative conclusion (all versions of (3)), via a normative cognate of one side of the conditional or biconditional (in all versions of (2)). That is, this sort of argument avoids the problem, because it is completely descriptive (using the necessity version just for illustration):

1. If A's were (able to) φ, then A's would have access to research.
2. A's are (able to) φ.
3. Therefore, A's will have access to research.

Or this, because it is completely normative:

1. If A's should be (able to) φ, then A's should have access to research.
2. A's should be (able to) φ.
3. Therefore, A's should have access to research.

The descriptive version is, of course, useless, as it's the normative conclusion defenders of TOA want (A)(ii)-style arguments to endorse. The only difficulty I have with the normative form is the inherent problem in establishing (1), whether in necessity, sufficiency, or necessary-and-sufficient versions. It's very hard to connect oughts together, with one as condition for the other, except in fairly trivial cases (e.g., If I should be a square, then I should be a rectangle).

We can't build a version of the (A)(ii) argument scheme that employs a conditional with a normative consequent and descriptive antecedent. That is, we could rebuild the (A)(i) argument scheme using a conditional:

1. A's have φ'd research.
2. If A's have φ'd something, then A's are entitled to access to that thing.
3. Therefore, A's should have total open access to research.

(Keeping in mind this employs TP.) The conditional here has a descriptive antecedent -- A's have φ'd something -- and normative consequent -- A's are entitled to access to that thing (i.e., A's should have access to that thing). That works because the argument is moving from a descriptive condition to a normative conclusion. (A)(ii)-style arguments don't run that way, though. They go from some condition that should -- but need not -- hold (i.e., some state that would be better than the status quo) to a normative conclusion. So, the arugment is normative all the way down.

Therefore, there is no really satisfactory form of (A)(ii)-style argument. All forms involve TP, and, additionally, the sufficiency form is fallacious, while the necessity and necessary-and-sufficient forms are question-begging against the status quo. Overall, then, (A)(i)-style arguments are superior to (A)(ii).

Moving on to (B)(i)-style arguments. These will look something like this:

1. If A's had access to research, then A's would gain ψ in order to φ.
2. A's should be (able to) φ.
3. Therefore, A's should have access to research.

The same problems present themselves: this affirms the consequent, it derives a normative conclusion illegitimately, and it employs TP. If we try to avoid affirming the consequent, we will, as with (A)(ii), end up with a form that begs the question against the status quo. If we try to avoid deriving a normative conclusion illegitimately, we end up with a premise that is of unclear truth, at best. And, as always, we're still using TP. Since (A)(i) arguments only use TP, they are still superior.

Similar remarks apply to (B)(ii)-style arguments. (Really, do I need to spell it out any more?) At the end of the day, (A)(i)-style arguments look the best possible. They do not commit obvious fallacies, beg more than one question, nor try to proceed from what would be good to what should happen. Instead, they draw a clear connection between what some class of people, A, has done (φ) and access to research. That is, they are (as is typical with entitlement arguments generally) backward-looking and not (broadly) teleological.

The problem, though, as said is TP: the totality presumption. Why does entitlement get us to TOA rather than IA? I can see two possibilities. First, arguments in favour of IA over TOA and the status quo fail. Thus, TOA is a fallback from the failure of IA. This is unstable, though, as it depends on the (lack of) creativity of defenders of IA. Second, IA is not stable in its own right; it will, inevitably, tend either to TOA or the status quo. But this is just the black and white fallacy. So, for TOA to be preferable to IA -- that is, for TP to be true -- it must be that there are no good arguments for IA.

Are there? That will be my question for next week.