[Tichy]

I just see that some billion dollar companies that got big through the web or computing are already going downhill again. For example Yahoo and Nokia (not sure about Microsoft).

For business management, off the top of the head I would ask: what devices will be used to run it in the future? Will desktop software still exist? Will everything be in the cloud? How will data enter the system?

One scary thing about the internet is that for a lot of things, a single company can cater to the whole world. That eliminates a lot of opportunities and competition.

[einhverfr]

There are changes coming. Nobody can dispute that. However, in a lot of cases, desktop software is not going to go away, at least for some definition of desktop software. A standard keyboard and screen interface is just too efficient for a lot of things. You aren't going to select, review, and adjust, and approve payment of 5000 invoices on your iPhone, and really, that's pushing the limit of stateless web interfaces too. You could have something like node.js or the like, but state really has to be handled or else you end up with major performance limitations.

Data entry devices however will become more diverse. There is no question about that. It used to be you had store-and-upload portable data terminals and desktop computers. Now you have both of those, plus the possibility of higher-end PDT's with real-time connectivity running embedded Windows, or the like (there is a real market here I think for Android-based PDTs but I haven't seen any on the market yet). These days, a PDT is kind of like a PDA, but typically more robust/rugged and often with additional industrial I/O capabilities, such as an RS-232 port which connects to undecoded laser barcode scanners. Doing this sort of thing on your phone isn't there yet. With a laser scanner, one can scan barcodes and take inventory fast, with barcode software on an android phone, your light limitations and low speed of processing make this problematic.

However, now we are seeing phones be used for some things. A worker may show up to a construction site and start entering time and material cards on his/her phone.

So what we see here is that each development is bringing diversity. The older layers don't really go away as much as one might think--- you can still buy PDT's which only store data to upload over a serial port and have digital LCD monochrome displays.....

Anyway our approach is to loosely couple the parts which will change only slowly (accounting logic) with the user interface which will be web- and desktop- based. Providing web-services and discoverable db interfaces makes integrating other devices easy.

We made these decisions for reasons other than being future-proof but that they help there is a nice bonus.

[einhverfr]

One more thought about old layers.

It has only really been about five years ago that we could get rid of terminal-based access (as in a virtual tty) to point of sales. A lot of people demanded something simple, and highly optimised for rapid data entry. The nice thing about a text-mode display is there is no temptation to take your hands off the kb and move to a mouse. KB and barcode scanner are all you need.

Now we have a web-based interface with a fair bit of automation, but we have to have a trackball in retail environments. However we make sure there are keyboard shortcuts to all the right things and that people are trained on these.

Of other businesses I know of, the last ones moved from text-only POS programs to graphical ones really only two to three years ago.

So things change remarkably slowly. The challenge is to accommodate a flexible workplace.

The web works for many things but for POS software, it is rather sub-optimal. Controlling serial port pole displays from a web page is a bit kludgy....