There are several different issues here:
Q1 Whether or not simple self-replicating systems can be considered life.
Q2 Whether or not it helps making a testable "definition" of life.
Q3 Whether or not life emerged as protocells.
All these Qs are non-trivial, so there isn't any consensus on this.
To take Q2 first, many astrobiologists find it unnecessary to define "life". McKay find 3 senses of definitions, none of them really helpful:
"What is life? This is a question that is often asked and typically confused.
The confusion starts from the several uses of the word “life” in English. There are at least three usages as exemplified by the following questions:
1) Is there life on Mars?
2) Is there life in this organism?
3) Is life worth living?
The definition of “life” in these three usages is quite different. In the first case, life refers to a collective phenomenon, in the second case it refers to the ability of an individual organism to metabolize and grow, and in the third case life refers to the history of activities that an organism undertakes. The first two usages are of direct relevance to astrobiology."
"Many commentators hold the view that an effective search for life on other worlds requires that we first have a concise, agreed on, definition of life. This is not the case. Along this line, it has been suggested that once we understand life we will be able to produce a completely mechanistic and predictive theory of life. The example of water is sometimes used. Water is simply defined as two hydrogens joined with one oxygen.
However, life is not a simple substance like water, rather it is a process, more like fire than water. There is no simple definition of fire. If life is like fire then even with a complete mechanistic and predictive theory of life we may still not be able to define it in any simple closed form. The search for life on other worlds can be based on what life does rather that its definition. One of the things that life does is build up large specialized molecules, such as DNA and proteins."
[
http://www.astrobio.net/news-exclusive/life-tricky-often-confusing-question/#sthash.8RvkQ7QK.dpuf ]
My own opinion here is that when people drag out the NASA definition (metabolizing self-replicator) they often, if not always, do so to try to identify an individual organism as a motivation to define life. (Say, how to quickly identify any extant cells on Mars.) But it isn't a good motivation, evolution as the process of life acts on whole living populations, so populations (of sets of alleles, in modern cells) are the vehicles that does that life do. As McKay says: "It is often pointed out that the definition of life as a system capable of evolution implies that single, isolated individuals not of child-bearing age are not “life.” This is nonsense and confuses the first and second cases of “life.”"
Now Q1 becomes slightly unhelpful even, except for fast and easy categorization. As atyy notes, it is a matter of [unhelpful] convention.
Q3 is the interesting part. It goes back to this:
stedwards said:
If you had the RNA without the membrane, the RNA would become dispersed. However if there were a single self-replicating strand this wouldn't be a problem; no membrane required. Is there a reason a single strand wouldn't work?
Well, it would be very unlikely that a variation of strands would immediately hit on a self-replicator. Indeed now that people have managed to make non-chiral, geophysically produced strands self-replicating, they find that so far only cross-replicators can do this and that they are shorter than self-replicators. I.e. a left-handed RNA replicates all the right-handed strands due to better specificity, and that includes the right-handed replicator that reproduces the left-handed pool of RNA strands.
Such a pool should dilute the cross-replicators eventually even if they grab local strands. So a membrane is likely necessary. Dunno if it has been shown, but I have seen the claim somewhere.
But I hear it is worse, such a pool would tend to shrink the strand lengths until they can't reproduce. To chemically select for longer strands, and start the "selection" part of life's processes, you need a hydrothermal vent that do metal atom PCR in cracks or pores of ~ 0.1 mm size (or larger). Thus far they have managed to do DNA with the usual PCR enzymes as proof-of-principle. So a whole alkaline hydrothermal vent is likely necessary to start chemical self-replication.
Picking that apart a Hadean alkaline hydrothermal vent had metabolism* (which produces the nucleotides), inorganic cells with membranes (that support metabolism and chemical replication), and self-replicating chemistry.
Last week I read Martin's and Lane's latest (?) paper where they, quite convincingly I think, show why the last universal ancestor needed to be dependent on an alkaline hydrothermal vent in order to evolve chemiosmosis and then become independent of the inherited pH differential. It seems simply impossible for a free-living protocell to have evolved chemiosmosis, while there is a mechanistic, even selective, route to it in a vent. Possibly protocells evolved chemically elsewhere and then infected vents to make the leap to robustly metabolizing cells. Else it seems easier that life evolved around vents in toto, and spread between isolated such as inert 'spores' at times.
This ties in with bacteria and archaea having evolved the different flagellum respectively archellum afterwards, as well as different root metabolisms, membranes and pumps for chemiosmosis.
[Personally I think, if M&L et al are correct, that the LUCA had pili for movement inside and outside the vents, and was not as tied to the inorganic compartments as they propose. The archaea archellum evolved from a, shared, pili ancestor mechanism.]* Shown by a neat proof-of-principle paper a few weeks ago, how such vents had to produce simple organics and possibly why some seem to do so a little bit even today.
