Empathy means understanding other people's experience, not using oneself as the only use case.
My argument is that we are predisposed, perhaps inevitably, to have more empathy with people like ourselves, and so removing empathy (rather than trying to fix it) may in some cases be the best way to get a fair result.
Basically, as far as I can tell, you’re advocating pulling the “politics” out of “policy”, but that’s really damn difficult, perhaps impossible. Just stating it as a goal is not useful, without some kind of concrete proposal.
The people in charge are always going to advance their own interests, and there’s no way anyone has figured out to put a completely impartial and dispassionate authority in charge of a government.
The way the US does things (ideally; in theory) is to build a system of checks and balances in the government, and try to spread the power widely enough in the population (e.g. via elections of local representatives who meet together to make policy) that any particular group is unable to exert their will without some measure of consensus and cooperation from everyone else.
The problem in practice is that some groups have gained so much power and wealth, compared to everyone else, that they can game and coopt many or most of these institutions; meanwhile, other groups have so little power or visibility that they are effectively unrepresented.
I guarantee you, if you build an institution designed to be “empathy free” — for instance, some board of highly trained unelected technocrats — to make decisions, their process will be analyzed and gamed to the extent possible by the super-rich.
So, from whence comes individual rights? The usual argument is, "imagine if that were you!"--that's empathy. More seriously, we guarantee individual rights for others because that means we have them for ourselves.
This leads to my second point: I think the reason empathy even is a thing is that we are inherently selfish, and for anything to be really done for the "greater good" individuals need to either have a stake in that greater good or they need to trick their selfish selves into thinking they are actually in that situation too, which, again is empathy.
Empathy works for the most part. It sounds like the author is trying to fix something that isn't broken. I'd say the current problems in Washington are in fact a lack of both listening to data _and_ a lack of empathy. I think more empathy would be beneficial, not detrimental.
That's a viewpoint problem in itself. Liberties and individual rights should never be added because they are a 'net benefit'. I see the converse as being true, liberties and individual rights should only be removed if they are a significant detriment to society. In your example of gay marriage there is no data that it is harmful, therefore no reason to prohibit such actions.
Of course proving something is harmful is not a carte blanche to prohibit most actions. For example our 'war on drugs' motivated as doing good against the evils of drugs as turned into a rather harmful system in itself. Other systems that treat drug users rather than punishing them have much better outcomes.
Theoretically, I think a good first step would be to stop evaluating policies deontologically and start looking at results. There are so many laws in the books with good intentions but terrible outcomes, and they're all justified based on empathy and related emotional responses.
A rather pedestrian example, I was having a discussion about transit planning with a group of people. Despite the less than enthusiastic reception transit gets from typical middle class white people, there's still a sense that the system is biased towards them, and doesn't serve communities of color/etc as well. The modern urbanist sensibility here is to have community engagement, neighborhood meetings, etc. All very grassroots. But who shows up to these things? Older white educated people mostly, usually with negative opinions to share. In this discussion people were very concerned about how we could get more representation from other communities, how we could hold more of these meetings in relevant neighborhoods, at times when working people could attend, etc.
The underlying logic of this participatory model is basically empathetic. People individually express themselves so that they can influence each other and the decision makers. They do this through empathy. The decision makers are generally more white, more privileged, more educated (and we can't expect to recruit more uneducated urban planners!) These people will be more receptive to perspectives coming from people like them. The response is often to increase the participation, be more inclusive, broaden the perspective. But minorities will always lose out, increasing participation doesn't disrupt that empathy bias. You can have more meetings in neighborhoods with minorities, and the white people even in those neighborhoods will still out-influence the minorities.
Can we fix the empathy problem in policing by getting more minorities on the force? That has not seemed very effective. Can we create better services for minorities by listening to their needs and responding? Eh, the best I've seen is people feel listened to, which is too often the substitute for real change. So I wonder if instead we stop relying on empathy in these circumstances, as emotionally attached as we are on the idea of empathy.
Going back to the original example, if we want to design a decent transit system we need to ditch the community feedback and go to data. In that model everyone is counted the same. Hidden bias isn't enough to create a model where a black person as 3/5 of a white person. And of course people still select models based on feedback, and that feedback will be biased, but it's way better than what we have now. Instead of trying to fix bias in order to fix social injustice, I think we'd do better constructing systems that remove the effect of bias in those places – which is the opposite of what most social justice advocates intend when they suggest we address the "root" of the problem.