Brendan wrote:
Schol-R-LEA wrote:
In any case, the paid upgrades were mainly a way to offset the production costs of the CD distribution - contrary to what Brendan said, upgrades were always sold at a loss.
Currently the retail price for Windows 10 is $150 to $250 (depending on version). If you think that's "sold at a loss" then you're a fool.
I suspect I am not the one being a fool here. Seriously, it isn't as if retail sales of Windows presents a significant part of their market (less than 2%, if memory serves, which is comparable to Linux's desktop market penetration - that is to say, negligible). Having it available for sale at retail at all is mostly a matter of PR. The price is just high enough to keep people from buying up large numbers of copies and reselling it at, say $150-250 (the arbitrage I mentioned earlier).
Brendan wrote:
You're all misunderstanding me.
Sometimes people upgrade the OS without changing their hardware at all. Sometimes people upgrade their hardware without changing the OS at all (e.g. just re-installing the OS they already paid for on a new computer). Sometimes people upgrade both at the same time.
With "rolling release" everyone can pay for Windows 10 once and then upgrade their hardware whenever they feel like it, and never pay Microsoft anything for any other OS ever again. That means existing customers will be paying $0 for "full OS", $0 for "upgrade only", $0 for OEM copies, and $0 for enterprise licences. It means the only people paying Microsoft for any OS are new customers, which is almost nobody.
Sure it's more convenient to have someone else install the OS when you buy a computer (unless you're "enterprise" and have admin/maintenance staff who will do it, most likely using a "slipstreamed from network" approach if they're large enough). Most local shops will do that for you already - you just ask when you buy the computer and give them your Windows licence key (unless you have the "locked to one specific computer" OEM version).
Brendan... seriously? 'Local shops' represent a tiny fraction of actual retail computer sales. So do custom integrators and self-builds, at least outside of the gaming community. Something like 95% of all computers sold in the retail market are from major vendors such as Dell or HP. I doubt that Microsoft is really hurting to go after the other 5%, especially since those will almost invariable buy Windows anyway - even Mac and Linux users often end up using a dual-boot setup, or running Windows virtualized under their primary OS, though again, the perception of how common that is gets distorted by the number of articles and forum posts about it.
In fact, the retail market in general is only about a quarter of the total PC market. The majority of PCs are sold for business, something which I am sure that you realize, given your target market. Of those, most are on the desks of some 9-to-5 worker who can't even install their own software (as in, even if they were allowed to, they wouldn't know how to). A small handful are in kiosks and such, but that's not really a notable market. The remainder are servers, which in the past usually meant a regular Windows box (which might or might not be running a server version of the OS) sitting in a closet somewhere, overheating and collecting dust; today, more often than not it is a rack mounted 'cloud' server which isn't even the property of those using it, with the stakeholders not realizing that this basically amounts to some company such as ShillFarce holding their data and software for ransom.
/me re-reads what I just wrote Man, I need to calm down. Talking about that particular company is still pretty triggering for me, though if you've ever had to work with that mess of a CRM you probably understand why.
Anyway, the point is that the x86 market is like an iceberg. The things that people talk about most are the small, visible part of something much larger, something that is a lot more matter-of-fact and pedestrian but far more profitable.
Brendan wrote:
Schol-R-LEA wrote:
As a related aside: shockingly, a lot of ATMs appear to run Windows NT 4.0 or even Windows 98, which is insane; I mean, why would you use a stock OS for that at all, never mind one so old and insecure? But as I have said before, the cost of security is a lot higher than that of insecurity, so they just don't care.
Sure, it's shocking that a sealed metal box connected via. a restricted network connection (with no ability to install software, nobody plugging in arbitrary devices, no web browser, nobody downloading pirated stuff from dubious sources and nobody clicking on "hot singles near you" adverts on porn sites) needs any of the security that Windows 98 provides. About the only security they do need (other than "physical security" - the tamper-proof enclosure needed to protected the physical cash that the machine dispenses) is a firewall that blocks everything that didn't come from one specific IP address combined with encryption to make sure packets can't be forged or interceded.
OK, point taken.
Brendan wrote:
Schol-R-LEA wrote:
But the real reason is that Windows itself is falling by the wayside, at least in the consumer market. The majority of consumer users aren't running desktop systems - or even laptops - at all now; most of the things which people would want those for are now done on smartphones, tablets, and smart TVs, even if they aren't suited to it, simply because they need a phone, want a TV, and want to be able to watch TV on a smaller screen when away from home, but they really don't see a need for a dedicate computer even if the lack of things like a mouse and physical keyboard is a hassle (which really only matters to writers, students, secretaries, and programmers). Most of the current Android and iOS users (especially in China, India, and Africa, which are the biggest markets for them, though it is also true to a lesser extent in established developed nations) have never owned a Windows system, and probably never will - in fact, quite a few have never even seen one in person.
Everyone that never used desktop/laptop systems before smartphones/tablets existed continues to never use desktop/laptop systems now (and just use smartphone/tablet); and everyone that used desktop/laptop systems before smartphones/tablets existed continues to use desktop/laptop systems (in addition to smartphones/tablets).
I think you'd be surprised by how many consumer users who had PCs in the past have stopped using PCs. Though as I have said, it isn't as if the retail consumer market was a major factor to begin with - it never really was one for PCs, actually, as prior to around 1988 or so most home users were still using 8-bit machines (or 32-bit home systems like the Amiga or the ST, though unlike in the UK, Europe, Japan, and presumably Australia, that wasn't especially common in the US), which also was around the time when small businesses in general started to take to the idea of using computers.
Brendan wrote:
There's only 3 things that changed - people aren't upgrading hardware as often as they used to (and therefore sales are growing very slowly for desktop/laptop), fools started making incorrect "number of sales == number of users" assumptions, and fools started comparing "rarely replaced, saturated market" PC sales to "frequently replaced, unsaturated market" smartphone sales. Note: I'm using "Occam's razor" a little here - it's possible that ARM vendors encouraged this stupidity to increase "ARM vs. 80x86" publicity despite the fact that ARM and Intel are different relatively unrelated markets).
I think you are missing just how big this shift actually is. It isn't really about numbers at all; the fact is, they
aren't related markets, as you said yourself. The difference is that the retail PC market isn't just slowing, it is
declining - because most of the people who were using PCs in their homes didn't really want PCs, they had them because they were the only reasonable way to do certain things. The only reason things like 'WebTV' didn't pick up before was because they were so terribly implemented, and because trying to surf the web on an NTSC, PAL, or SECAM television made your eyes hurt.
Things like smartphones, tablets and smart TVs give consumer users what they actually wanted from PCs all along - and frankly, they don't really want all that much. Most people want just three things out of these devices: to buy stuff from places like Amazon, Alibaba, Wish, and eBay, watch streaming video from YouTube and Netflix, and play casual games like Candy Crush. That's about it, sad to say. The truth is, most mobile devices are vastly
overpowered for the purposes they are put to.
Brendan wrote:
The fact is that (for PC in general) the only significant change (that can't be attributed to reaching "product maturity") is that a lot more servers are being sold to datacenters/"cloud" than ever before; partly because smartphones are too weak to do any real processing and need to offload work elsewhere, partly because of all the server consolidation/virtualisation hype, and partly because of the increase in marketing companies tracking and analysing users ("big data").
I would be inclined to blame that on things like Node.js instead, though the volume of easier-to-use frameworks probably is related to the demand for webapps aimed at the phone market, so.. meh. Oh, and on spyware, which isn't quite the same as what you were describing about tracking (being more of a source for the data than the processing of it, and yes servers do get infected with malware often enough for it to be significant), though the line between them is a very fine one indeed, and one often crossed.
As for the processing power of smartphones and tablets, that's pretty much inevitable - the main limiting factors there are heat dissipation and battery life. Neither the software nor the CPU architecture being used for them can really do much about those - the main reason ARM came to prominence there is because it is somewhat easier to tune for that than most other ISAs, but even that is relative. An ARM implementation that could match the desktop x86 chips would draw about the same amount of power as the x86s would, perhaps a bit less but not significantly so, and have about the same heat dissipation requirements.
The same applies to laptops to a lesser extent, but that, too, is a different market - and contrary to the general impression people might have, not an especially large one, being primarily 'people who would be happy with a tablet except they need to type a lot more than would be comfortable on a virtual keyboard or a portable tablet keyboard'. Any hardware market where IT consultants (particularly of the office-free hipster variety, like I had been at times, as more conventional consulting developers generally use their client's office hardware or the desktops in their own offices) represent a significant part of the user base is by definition tiny.