And software costs... this was also the era when every "application" (think accounting software: accounts payable, payroll, accounts receivable, word processing) was $495 per application. A small business could easily pay $2-5K for basic software to run their business. And then, of course, it was a nightmare to setup and use and almost impossible to pull off without a "consultant".

But VisiCalc -- it was such a game changer. A totally different way of using a PC that enabled the "ordinary" non-computer person to become an order of magnitude more productive. I think Lotus 123 was the pinnacle of golden era of keyboard-driven spreadsheets, but it was only an incremental improvement over VisiCalc. The journey into the abyss began with Lotus Symphony.

I do like Excel and use it on occasion. But I pine for the days of lean software that did one thing exceptionally well.

it was a lifetime licence though. you can probably still run it today.

Unless you keep running every part of the hardware / software stack, which works only until some piece of hardware wears out.

exe34 15 hours ago | root | parent | prev | next [–]

> $495 per application. it was a lifetime licence though. you can probably still run it today.

Also:

that's where we disagree: you don't usually need to stay current. as long as it does the job, it's current enough. if there's new features available that would add value to the business, then you have a business case to buy a new license. 95% of software update haven't really added any value since the early 00s.

-----

Maybe I misinterpreted but the implication is I could run something from the early 00s without changes which--while true in some cases--I wouldn't do in general.

Yes, you did. If he meant that, he wasn't thinking. Some piece of your hardware won't work with the old software, unless it's also old. You get a new printer and there are no drivers for the old OS. The old software won't work with the new network. You want some new app and it doesn't work with the old software. Etc. Etc.

We ran into this in Google Patent Litigation all the time. You have to have old everything to run old software, whether your license is still good or not.

But I think we agree that paying for maintenance only, and no new features, is fair. Of course, most vendors don't want to do that; they want to shove new "features" down your throat.

I've also known customers who have old software, e.g. for test systems, just running on old systems who basically don't breathe on the ancient systems. For years, United's entertainment system would also sometimes reboot to a 20+ year-old pre-Red Hat Enterprise Linux kernel but that's not really a critical system.

I always think of the John Mellencamp line, "That's O.K., I knew this would happen

But I was hopin' not today"

https://genius.com/John-mellencamp-im-not-running-anymore-ly...

They sat on their laurels with 123 v2 for much too long, and competitors surpassed them. v3 wasn't enough more to catch up.

Enable OA's spreadsheet module[1] was certainly head and shoulders above 123, offering real 3D capability (where what Excel views as multiple independent sheets could be addressed directly, so you could do things like have a layer for each month of an annual report, with the topmost being as @SUM() over the column of montly numbers below it), much richer set of functions, and integration with a database and word processor.

And there were others, too. I liked Lucid-3D, which wasn't really 3D but kinda fractal, where you could set up any given cell to drill into a sub-spreadsheet that was used to calculate a single value that would roll up into its parent. And Borland had a competitor, but I don't recall anything about that one.

[1] During college I had a part-time job working on this, on the testing team.

One cool thing we testers did to keep ourselves entertained was to build a spreadsheet casino. Each of us took on a given casino game to implement via spreadsheet macros. My game was blackjack, and it supported the full range of features: multiple decks, double-down, insurance, and all that.

Another guy did craps, the result of which was that we found a subtle bug in the app's random number generator. He set it up to play itself automatically and left it running overnight. When we came back in the next day, he was rich; that's not supposed to happen. He ran the test again the next night, and same thing. The bias in the RNG was causing rolls of 11 to happen more often than they should have.

Pyspread has some interesting features along those lines.

DOS was $40. CPM/86 was something like $186. Which is why everybody bought DOS instead of CPM/86.

My first computer (other than single board CPU projects) was an H11. You can see it on my profile:

https://twitter.com/WalterBright

I'm still sitting in the same chair. The desk I made in wood shop in high school.

Some of the early PC Magazine utilities were cool.

Early in my career, I read some books about it, including a few about programming for the 6502 processor, and later, Peter Norton's "Assembly Language Programming for the IBM PC". I thought the latter book was very good, just like his Norton Utilities and other software.

I did a fair amount of hobby 6502 assembly language programming on the Commodore 64, and a bit on the BBC Micro. The Beeb had a cool inline assembly feature, using which you could write 6502 assembly code inline in the middle of your BBC BASIC code, just by wrapping the assembly code in square brackets. There was no need of an explicit assembly step. The inline assembly code would just run as part of the running of the BASIC code.

I used to daydream about the equipment in the ads like other kids did with cars in car magazines

I believe you, ha ha. I used to do the same, with both PC Magazine and BYTE.

And with DDJ and CUJ too. :) (Dr. Dobbs Journal and C Users Journal. Both of those had high quality articles on software topics, back then, and for many years.)