Spartan159 said:
if retro tech computers are half the cost and weight per step backward, would programs be less expensive after their TL of introduction? For instance, Expert/2 is available initially at TL 12 for Cr10,000. Would it cost less at TL 13? Less still at TL 14? Only talking about the price, not bandwidth. I have full confidence in code bloat.
Software for older computers needn't be intrinsically less expensive, although you could argue that there's an eventual trend to commoditisation. For example, once open-source O/S software like Linux or BSD became widely available on commoditised hardware, it created a significant downward pull on the value of operating systems. In fact, the real value of an operating system is that you can run your applications software on it. One might argue with some justification that this is the only reason the computer industry puts up with Windows. Several vendors still make traditional mainframe systems and have customers willing to pay large sums of money to run their portfolio of applications. While reliability and throughput play a factor in this this could be achieved with more mainstream kit in many cases but the incumbent software has such a value and high replacement cost that it's still cheaper to pay for the hardware.
You can see similar phenomena in aerospace software; the CPU in the main avionics computers on a F22 dates back to the 1990s. It's about as powerful as a Playstation I. The case to rip and replace the avionics systems with more modern hardware just isn't there because the R&D and certification costs would be prohibitively expensive.
Computers can also maintain a large degree of backwards compatibility if desired. For example, modern IBM mainframes can still run binaries written for a System/360, which was first made in 1964. Even the humble PC has a fair degree of backward compatibility available through use of VMs and (although more spotty) supported legacy APIs.
Software costs are largely going to be a function of market forces. For software designed to run on a starship there is only going to be a finite market dictated by the number of starships. The software isn't going to be useful outside this market. If the number of starships increases significantly then the vendors may be prepared to reduce the price of their software in order to get access to a share of a larger market. From this perspective, obsolescence of the software is less of a force than the size of the market (after all, plenty of folks still pay top dollar to run SAS, SAP R/3 or Oracle).
The other force at play would be commoditisation of the software, an example of which can be seen with open-source systems. In some markets (e.g. software development tools), open-source is king, and there is relatively little appetite in the market for any benefits that commercially extended versions could bring. Most vendors of proprietary development tools are either slowly making their core tooling available in this mode (even Microsoft) or receding into niche markets (e.g. Embarcadero's purchase of Delphi, which largely supports existing applications software written on that platform).
What one could perhaps see is a secondary market in certified avionics systems running commoditised, older versions of software (especially if this software was developed by state actors and released as a public service) and more high-tech, cutting edge kit running modern software. The former might be quite cheap and the sort of kit you might find in remote areas, made by local vendors based on older I.P.
