For years, retrogamers have wanted new games for their old unsupported systems, especially the venerable 8-bit Nintendo.
Hardcore hackers have approached the problem by creating new games, or rediscovering lesser-known ones. People usually play these on an emulator, but it just isn't the same.
For
the most hardcore, there is the option of creating a new cartridge using an old one as a "parts donor". Also there are expensive (usually in excess of 100 USD) flashcarts.
I envision a cartridge every homebrew hacker would soon be able to afford. The solution is to use a simpler and thus cheaper design.
Typically, a NES cartridge consists of a circuit board with contacts on one end, housed in a plastic casing. Mounted on the circuit board are various components, connected to the contacts. Usually these consist of 2 ROM chips, one for the program code, one for other data, a mapper chip, which performs memory paging, and the 10NES chip, which acts as the key to the (identical, but in lock mode) 10NES chip within the console. Some cartridges contain extra RAM and/or other chips that enhance the capabilities of the console.
The first problem we have is the 10NES lockout chip. There needs to be one in the cartridge or the console will not accept it. Hackers usually work around this by either using a "donor" 10NES chip from a original cartridge, or by disabling the 10NES chip in their console. Pirate company Color Dreams worked around the 10NES by temporarily knocking it out with a power surge. Tengen's solution was to create their own copy of the 10NES chip after reverse-engineering the original. Later, they integrated their "rabbit chip" as it was called, into the mapper chip.
Many people don't want to mod their NES, so we shouldn't have to disable the chip. Color Dreams' method doesn't work on all consoles, and is unreliable even on consoles it is known to work on. In this day and age, we have cheap, reprogrammable, general-purpose microcontrollers, so I think Tengen's method is the way to go. Much about the function of the 10NES chip is known, so recreating it should be relatively easy. Also, I don't like the idea of having to destroy an old cartridge to make a new one, because no matter what game it is, it's a piece of history. Since we will
need a mapper chip anyway and any general-purpose chip that can be a mapper chip is overkill for a 10NES chip, one chip should be all we need for this.
So now we need the program and data storage, which is accessed by console through the mapper. Usually there are two EEPROM chips for this purpose. With a special mapper chip, however, only one large chip is needed. A single NAND flash chip would be used for this, because these chips are convenient and are now cheap.
But where would the mapper/10NES chip's firmware reside, you ask? Some general-purpose microcontrollers nowadays have their own flash memory and RAM integrated. But if we wanted to use a cheaper chip, an area on the NAND chip could be reserved instead. Since NANDs are rewriteable, an area could also be reserved for game saves (SRAM). No battery-backing would be necessary.
Some I/O pins on the mapper/10NES chip could be freed if shift register chips are used. This could allow for a cheaper chip, or for the chip to also serve other purposes, for example it could act as a DSP or co-processor to allow for more advanced games. More importantly, USB upload connectivity could be added.
Casing is the easiest part. Homebrew plastic case-making has been pioneered by folks such as Ben Heck, and is now a skill known by many. The simplest method would probably be to paint and label some plexiglass, and glue/screw together.
This is just the basic model. Other features such as additional RAM (which some games require), multicart functionality (this should be possible just by using a larger flash chip and a different mapper/10NES firmware design), or an extra microcontroller for more advanced games, could be added if desired. Perhaps even debugging features could be added. Somebody enterprising enough could make good money by selling these.