Journals like Nature and Science have impressive impact factors. How and why did these top journals become top journals? Why are they able to sustain their statuses?
All journals that have a high standing have the standing because of the support of the community. If the community loses interest, the journal will drop in the ranking. The top journals have therefore attracted authors for one reason or another. The editorial staff of journals try to maintain this status by making sure the work published there is of good quality and will be cited. It is thus not impossible for new journals to attain high status as long as authors provide the necessary papers. To this mix, we now also add the impact factor and other bibliometric factors. They matter now but have not been the driving factor for making the older journal what they are today.
I work with a more modest journal and I can definitely state that improving your impact factor is far more difficult than to drop in ranking. But, if your ranking becomes high enough (no specific number will be relevant since it varies between fields) a journal will be self-fuelling since many want to publish their material there and competition stiffens leading to a strong selection.
So the standing of Science and Nature, is part their long history, in part the hard work by the journal itself and in part the, now, need for authors to publish in as high ranking journals as possible since that is what forms the basis for most evaluations in academia.
I think this is analogous to "why is Harvard a good university, and able to maintain its standing as such?" A partial answer is that (1) it was founded a long time ago, and (2) it was founded by serious people. Given that, further serious people will tend to gravitate to the same institution, creating an inertia in the rankings.
A quote from The Crucible (set in the year 1692):
I am not some preaching farmer with a book under my arm; I am a graduate of Harvard College.
Suppose you start with a collection of journals and people who want to publish quality papers in them, who arrive over time. Suppose that each new quality paper is sent to a journal which is chosen at random, but where the probability of choosing journal X is an increasing function of the number of quality papers which have already appeared in X. Then you are dealing with a preferential attachment process and you will find that after a long time, most of the quality papers will be appearing in a few top journals and there will be lots and lots of mediocre journals with very few quality papers.
Naturally it's a very simplified model, but the same argument can be used for the sizes of cities, views of Youtube videos, distribution of wealth, etc. See Chapter 18 of Easley and Kleinberg's textbook for more.
One related comment-
There are some comments here that are equating high-impact factor with high prestige. I think this generally holds true, especially for people working firmly within the boundaries of a single discipline. For these people, of course, reputable journals are going to be more widely read and thus cited more frequently, and have higher impact-factors.
For people whose work are more inter-disciplinary, the relationship between prestige and impact factor are not so straightforward, because the size of the disciplinary audience can be very different.
For example, my work is interdisciplinary and lies at the boundaries of sociology, economic geography, management/organizations, and Asian studies. Journals in each these fields have different audiences, number of scholars, and thus, different impact factors. For example, Asian studies have a number of high quality high prestige journals that publish excellent papers, but because the size of the core audience, even the top journals hardly have impact factors that exceed 1. For management, however, because the field is very large, even journals that publish not-so-rigorous studies tend to have high-impact factors, easier exceeding top journals in social science/humanities. There are a number of journals that have impact factors over 5 or 6, and even mid-range journals have impact-factors around 3. Sociology and geography lie somewhere in-between.
It might be a natural tendency for people to try to publish in journals with high-impact factors. However, I would say, at least in my field, there is definitely higher prestige that will be attached to work in, say, top tier Asian studies journals (say, impact factor of 0.5-0.6) or a top- sociology journal (impact factor around 2-3) than a mid-tier management journal (impact factor of 3-4).
Journals gain their status mainly by being the first to offer publications in a new field and secondly by recruiting influential people in the field as editors.
Once a journal has a high impact factor people will want to publish there because authors themselves are judged on the impact factors of the journals they publish in. This means the editors can set a higher standard for acceptance. Since the impact factor is based on citation rates it then increases further. Most academic journals publish only papers and have very little editorial content, so this positive feedback mechanism is the main thing that maintains their top ranking. The mechanism operated even before impact factors were formally measured because people still knew roughly what the impact and standing of a journal was.
It is very hard for a new journal to get a good reputation because it takes two years for them to even be given an impact factor and this will remain low because they will fail to attract the best papers initially. A new journal needs to offer something different to succeed. They may specialize in a new field that does not already have a top journal, or they may offer open access for low charges in order to get going, but the established journals are always very hard to dispose.
The other factor that keeps a journal in the top ranking is its editorial board, but this is not because the editorial job requires their skills and knowledge. What the journal needs is a good supply of peers willing to review articles well and it is not easy to persuade academics to dedicate their valuable time to this chore when they don't get paid for it. The main reason they do agree to review articles is to impress the editors because the editors are influential people in the field who may help them get their next job.
Editors themselves take on the role because of the prestige of being an editor for a top journal and because they get an opportunity to identify reviewers who understand the field so that they can recruit them. This establishes another positive feedback that helps maintain the journals top spot. One of the few things that can destabilize a top journals position is the resignation of its most influential editors.
Whether this amounts to a good system for academia is very much open to debate. Most top journals are in the hands of big commercial publishers who understand how the system works and who have cleverly developed and promoted the journal impact system to their advantage. They make huge profits taking money from scarce scientific funds when most of the hard work in publishing is done by unpaid authors, reviewers and editors. Efforts by academics to change this usually fail because they don't understand how the system works, or because they dont have the time or funding to realize their ambitions. Another reason seems to be that governments and funding agencies like the big profitable corporate publishers so they tailor legislation to suit the publishers rather than the academics. Also the academic societies (APS, AMS etc.) who supposedly oversee the interests of the fields are themselves funded mostly through their journals so they have a massive self interest in perpetuating the system.