Why do publishers seem to thrive while developers fall by the wayside? Perhaps it is because publishers are obsessively focussed upon what is required for their own survival, while developers generally just want to make games.
Although I am not well versed in economics, it seems to me that all business is about purchasing lottery tickets. These tickets vary as to their price (the cost of investment), the chance of winning (the risk) and the payout (the return on investment). In broad strokes, the dominant strategy is therefore to minimise risk and maximise payout. Unfortunately, it transpires that every project has a certain risk attached, and this can never be eliminated, nor accurately predicted. The payout is similarly difficult to ascertain in advance. Once this is realised, the dominant strategy shifts: buy as many (low risk) lottery tickets as you can find.
Indeed, it has been demonstrated that the safest form of gambling is the stock market. Buying a diverse portfolio of shares is akin to buying many mid to low risk lottery tickets.
Publishers flourish (generally) because they work on the basis of portfolios of games (or in other media, of books, films, TV shows etc.) Since the risk and payout of individual projects cannot be known in advance, having a handful of lottery tickets is the safest approach. Several will fail, but if you have balanced your portfolio well enough, sufficient numbers will succeed to allow for an overall profit. The more steps the publisher takes to minimise risk (for example, by purchasing a license - sports, film, celebrity, TV etc.), the better the chance that the portfolio will show significant profit and the publisher will survive.
Developers wither (generally) because they tend to lack the resources to work on multiple game projects. They usually work on single game projects. They believe that this will work, because they know the game they are making is "good", by which it should be understood that they are making a game they want to play. In effect, developers expend great time and effort to manufacture a single lottery ticket - often one with high risk and low payout. Some get lucky. Most fail.
In the indie community, the situation is slightly better in that the cost of building a new lottery ticket is lower. Also, because the price of the tickets is lower, the returns can be much lower and still imply a profit - which is in effect a means of minimising risk. But still, the indie developers that succeed are those which either manage to get a portfolio together (PopCap I salute you!) or those that happen upon a "golden ticket". These latter souls often still fail in the end, because very few individuals are ever fortunate enough to find multiple "golden tickets".
The number of game designers who have founded multiple successful franchises can be counted on the fingers of one hand - even if you lost a few fingers in a farming accident.
Against this backdrop, I find myself reluctant to make games that cost more than a fairly modest amount to make, and I find myself constantly looking for new business models that might be applied to development to increase the chances of survival. But as development costs rise, the situation for developers becomes gradually worse unless they give up on making their own games and begin manufacturing products tailored for publishers (not an appealing prospect for many people). Intuitively, alliances could mitigate risks, but alliances between developers seem difficult to broker. Developers want to make the games they want to make. All other matters are secondary. Even their own survival.
I doubt this situation is going to change any time soon.