This is a timely contribution for an anniversary year for ALife, giving a perspective on the past and present states of ALife of a foundational contributor whose works are essential texts for anyone coming into the area. Some cogent criticism of present failings—such as inconsistency in peer review; lack of awareness of previous work, leading to wasted repetition—leads to some clear recommendations for the ALife community to steer its future direction(s). Essentially, the recommendation is for ALife to grow up and become a mature science, with a shared body of knowledge that can be incrementally expanded through systematic research programs.
This is an important perspective, shared by quite a few people, and cogently expressed. Of course, I must recommend it be accepted—even though I very fundamentally disagree with the diagnosis. Beer is hoping “to stimulate some necessary critical conversations”—OK!
Here is an alternative perspective. Artificial Life is not a field of study with shared, well-defined goals, and it should not be. It is a loose-knit, ever-changing community, looking for new interdisciplinary connections between any of the multiple fields of biology and any of the multiple fields of technology that can lead to new research ideas and, we hope, new research programs. On this perspective, the role of Artificial Life forums (journals, conferences, workshops) is to encourage the conception (in a very biological metaphor) of such new ideas, with new cross-fertilization breaking out from old constraints. While many will be stillborn or die in infancy, those that do grow up to be adult sciences will either be subsumed within preexisting, established fields or form their own established fields obeying scientific norms—typically moving outside the ALife umbrella in doing so.
Artificial intelligence, artificial neural networks, deep learning, genetic algorithms—they were each initially conceived in an ALife-like milieu; they grew up and survived as independent fields of research; they have left ALife behind. This, I suggest, is good and as it should be. Beer’s own research history, as he reports it, fits this pattern. He was inspired by attending the 1987 ALife workshop—a forum more for promiscuous, novel conceptions than for established, incremental science. For his computational neuroethology work, he found that journals other than Artificial Life offered better connections to more established science—fine from my perspective. Then, when he returned (2014 on) to illustrating issues in autopoiesis via Game of Life gliders, it is difficult to think of an appropriate place to publish—except, of course, in Artificial Life! From my perspective, he did just the right things and was well served by ALife.
I could suggest that if Beer’s current recommendations had enforced his “adult scientific norms” on the 1987 workshop and the 2014 Artificial Life journal’s publishing criteria, he would quite possibly have been uninspired by the first and rejected as out of scope by the second. Yes, of course, we want adult, incremental science—but it does not have to be at the expense of the opportunities that Artificial Life forums allowed Beer and others in the past.
Beer makes other specific criticisms that I acknowledge. Inconsistency in peer review is indeed a major issue. In part, this may be exacerbated by interdisciplinary crossovers putting reviewers out of their comfort zones; novel ideas can be mistaken for nonsense, and vice versa. I am in favor of making reviews public (whether named or anonymous) and advocate publication methods that support this, for example, the Open Peer Commentary format of the journal BBS (Behavioral and Brain Sciences) where a target article is published along with commentaries. Lack of historical awareness is an issue, though arguably, this matters less in the nascent stages of research than in later, “adult” stages. With the third stated problem of “lack of consistent progress toward any identifiable goal,” I can agree with the concern when stated thus. But I worry—and disagree—when just a couple of sentences later, this has subtly morphed into a shared goal, with an implication that it is impermissible for different groups to have different goals.
To recap the perspective on Artificial Life I am suggesting, forums that support the conception of novel interdisciplinary research ideas have very different requirements from forums that support adult scientific fields. Both types of forum are desirable. Artificial Life has a rather good track record in providing the first type of forum, and any attempt to eliminate such a role should be firmly resisted. The successful spin-offs have indeed thrived as independent adults—but we still need ALife for the as-yet unborn.