Hacker News Reader

Comment by petterroea | Hacker News Reader

A business man at a prior employer sympathetic with my younger, naive "Microsoft sucks" attitude told me something I remember to this day:

Microsoft is not a software company, they have never been experts at software. They are experts at contracts. They lead because their business machine exceeds at understanding how to tick the boxes necessary to win contract bids. The people who make purchasing decisions at companies aren't technical and possibly don't even know a world outside Microsoft, Office, and Windows, after all.

This is how the sausage is made in the business world, and it changed how I perceived the tech industry. Good software (sadly) doesn't matter. Sales does.

This is why most of Norway currently runs on Azure, even though it is garbage, and even though every engineer I know who uses it says it is garbage. Because the people in the know don't get to make the decision.

View on HN

I’d say, they are very good at making platforms and grab everyone lock-in. But they need a good platform first. Azure seems like the first platform that is kinda shitty from the beginning and did not improve much.

MBASIC was good and filled a void so it got used widely from the beginning. The language is their first platform. Later the developer tools like the IDE, compilers, still pretty solid if you ask me.

MS-DOS and Windows are their next platform. It started OK with DOS — because CP/M was not great either. But the stability of Windows sucked so they brought in David Cutler’s team to make NT. It definitely grabbed the home/office market but didn’t do well for the server market.

X-BOX is their third platform, which started very well but we all know the story now.

Azure is their fourth platform, started shitty and still not good. The other platforms have high vintage points but Azure may not have one.

Those are mostly end-user or hosting platforms you mention (and their problems), what really makes MS tick is the enterprise platforms.

Windows networks, Active Directory,etc. Azure is the continuation of that, those who run AD oftne default to Azure (that offers among other things hosted or hybrid AD environments).

Yeah those too, sorry I never worked with the MSFT stack in corporate, except for my first company when my IT knowledge was still minimum.

That's true for Azure, where contracts are signed due to free credits given over Office and Windows usage.

However, there is a reason why everyone uses Office and Windows. Office is the only suite that has the complete feature set (Ask any accountant to move to Google Sheets). Windows is the only system that can effectively run on any hardware (PnP) and have been that way for decades.

This is due to superior software on the aspects that matter to customers

People use Windows because Office runs on Windows, and Windows ran in any shitty cheap beige box. This is the whole story since the 1990's.

On hardware: it's because Windows has a stable kernel ABI and makes it very simple for hardware vendors to write proprietary drivers. Linux kind of forces everybody to upstream their device drivers, which is good and bad at the same time - DKMS is something relatively new.

But yeah, the NT kernel is very nice, the problem with Windows is the userland.

I use to think that too.

But if you really look at it the "comfort zone" problem isn't too big of an issue in itself that a few training workshops and brief acclimatization periods for other tool suites can't solve. Making accountants move to Google Sheets is actually doable given enough incentive; there really isn't a lack of features in Sheets against Excel so much as there is a difference of implementation. In fact, for many purposes Sheets and GSuite could even be the "superior software" if only one bothers to make good use of it.

The problem is more that companies hesitate to take the dive because they can't be sure any of the alternatives will stay stable in the long run. Google is infamous for abruptly shutting down applications and none of the other competitors have built enough of a repute yet to ensure longterm reliability.

Microsoft has been (and continues) riding on its first-mover advantage as an unmovable establishment for decades. It has worked out till now, but who knows till when.

The selling point of Excel is not the feature set, it's that people know Excel and are usually very resistant to learning something new.

As someone who’s compared spreadsheet feature sets, though: it’s also very much the feature set.

Well, in a way it is of course, because if your reference is Excel, then you want the feature set of Excel.

Or what specifically do you mean?

Sheets and Numbers are spreadsheets. Excel is an application platform and programming language that’s convinced people it’s just a spreadsheet.

VBA, PowerQuery, structured references, the newer formulae like XLOOKUP, dynamic array-spill formulae, map/filter/reduce/lambda, various obscure financial stuff.

Sheets and Calc don't have these.

The problem is that it encourages people to use excel for things that should never be in a spreadsheet in the first place. I mean if you're reaching for VBA, building complex PowerQuery pipelines, and writing nested LAMBDA functions just to process your data, imho you have outgrown excel. Just because you can build an entire solution in Excel because you already know the interface, doesn't mean you should...

Also, don't get me started on the newer functions such as XLOOKUP and Dynamic... Relational data belongs in a relational database. If you are joining tables and filtering massive arrays, you should be using standard SQL Arrays, it makes it so much easier to troubleshoot long term.

Windows is the only system that can effectively run on any hardware

...as long as that hardware is Intel-based (and a select few ARM-based boards nowaways). And the reason that it runs on all that hardware is because of Microsoft's business contracts with hardware vendors, not because of their software quality -- that's immaterial, as Microsoft generally does not write the drivers.

Compare the experience in Linux or Mac for getting some random no-name device working with Windows.

A lot of it is the fact that the OS has created a very complex yet consistent system of device compatibility that was completely absent from all competitors who are still behind on that aspect or alternatively the choice of kernel design architecture

It's been like two decades since I used windows on a computer I own, but I always had a way harder time getting hardware to work with windows than I have with linux. I still shudder when I remember trying to track down drivers from different vendors, while avoiding the malware they shipped with it versus letting it just work.

edit:

I just remembered when I first used CUPS to configure a printer in 2003. It blew my mind with how easy it was, and I think that was the moment when I decided to start using linux as my primary desktop. Pre-Novell Suse at the time if im remembering correctly.

This is in many ways a smart way to understand the problem, but it doesn't mean that microsoft contracts mean you're stuck with bad software. There are several verticals where Microsoft and Azure actually were smart and chose a better software product to sell on their platform than what they had in house.

One example is when they stopped trying to develop a inferior product to EMR and Dataproc, and essentially just outsourced the whole effort to a deal made between them and Databricks. Because of this I assume many enterprise azure customers have better running data solutions in that space than they wouldve had they gone with just AWS or GCP.

On the other hand, having worked for Microsoft on an Azure team, there are plenty of areas that critically need a rewrite (for dozens of different reasons), and such a solution is never found (or they just release some different product and tell those with different needs to migrate to that), where they keep on building what can only really be described as hot-fixes to meet urgent customer demands that make it harder to eventually do said critical rewrite.

The Databricks thing was a ploy. They then pushed Azure Synapse Analytics and forced all internal teams to stop using Azure Databricks. Synapse was half baked and then they are now pushing Microsoft Fabric which is even less baked.

About a year ago the whole situation changed and Microsoft started to push everyone to their own Data Engineering solution (Fabric) that back then was really half-baked.

But that also means that if you as a user/customer can make choices based on technical merits, you'll have a significant advantage.

An advantage how? Maybe you'll have one or two more 9s of uptime than your competitors; does that actually move the needle on your business?

The biggest expense in software is maintenance. Better software means cheaper maintenance. If you actually want to have a significant cost advantage, software is the way to go. Sadly most business is about sales and marketing and has little to do with the cost or quality of items being sold.

Why wouldn't it move the needle? Less time spent, less frustration, more performance, more resources focused on the business?

You massively underestimate the difference in employee productivity not having to fight user-hostile software every step of the way can make. Not to mention cascading cost savings and infra flexibility distancing yourself from Microsoft products can grant.

It will depend on each case and what makes the marketed solution inferior. If it's overly complex and you will save development time. If it's unstable you'll save debugging time. If it's bloated you will save on hardware costs. Etc...

matters less than we would like it to

after all startups/scaleups/bigtech companies that make a lot of money can run on Python for ages, or make infinite money with Perl scripts (coughaws)

and it matters even less in non-tech companies, because their competition is also 3 incompetent idiots on top of each other in a business suite!

sure, if you are starting a new project fight for good technical fundamentals

Most customers don't really have the knowledge needed to make choices based on technical merits, and that's why the market works as it does. I'm willing to say 95% of people on HN have this knowledge and are therefore biased to assume others are the same way. It's classic XKCD 2501.

A overly reductionist argument. They described any commercial software company because in the end, you sell or you die. Microsoft has incredible software people and incredible software that coexists with the shitty software people and shitty software.

Agree. You could say the exact same thing about Oracle, for example.

I think this is spot on. Everything at the R&D phase of a project indicates that an Azure service is going to work for the use case. I've been reading the docs and though 'wow this is perfect!'. Then you get to implementation and realize its a buggy mess that barely does what you wanted to do in the first place, with ton of caveats.

Of course that realization comes when you are already at the point of no return, probably by design.

My lesson was when European companies followed US tech into offshoring, and how quality doesn't play any role as long as the software delivers, from business point of view.

Especially relevant when shipping software isn't the product the company sells.

Finnish public sector is also heavy Azure user. Their common ethos is that modern cloud services(=azure) are in many respects more secure than on-premises data centers. In addition, they are cost-effective and reliable.

I mean, if you ignore all the heaps of impressive software Microsoft does ship, sure.

It’s been a while. The underinvestment shows. Across the industry as well.

Reply on news.ycombinator.com