Hacker News Reader

Comment by pickledoyster | Hacker News Reader

> Which is that Apple and Microsoft are the only real system vendors, corporations who can offer an entire integrated system and aren't just either single components or many different components packaged together, but without the interaction necessary to compete with Apple or Microsoft.

This is just a thought that ignores all the economies of scale etc., but what if monopolistic tech conglomerates were seen as a negative vs interoperable, modular systems? If that were the case, simply repeating US tech's blunders wouldn't be a true alternative, just more of the same with garden walls made of a different material.

View on HN

I think that is a question of architecture.

What is important that there is one company you can go to who does all of that for you. Running a government institution and having 20 different vendors to make your basic IT system work is a nightmare. That you can get all in one from Microsoft is one of their biggest strengths in the market and you must compete with that.

> there is one company you can go to who does all of that for you

While I understand what you're saying, isn't that surely the problem?

Putting all of your eggs in one basket may give you a nice vertically integrated system you can buy off-the-shelf with little effort, but then you're wholly dependent on that org for everything from the platform you're hosting your infra on, to the tools you communicate with and the software suite running on your workstations; having your org use _everything_ Microsoft might be easy, and a little bit spendy, but the moment Microsoft is off the table, you're left without an org.

Disparate systems from all over the place might very well be more effort, and also likely cheaper/free in terms of licensing costs, which you can then spend on creating jobs and/or contributing back to those systems. The larger your org, the more you'll save and the more you can spend on creating jobs, and more importantly, those jobs can be created locally.

Too much of the world depends on a few big orgs in the US with potentially different goals and values to their own.

> Too much of the world depends on a few big orgs in the US with potentially different goals and values to their own.

The solution is simple: build a business environment that would allow a home-grown alternative to have developed over the last 40 years.

Nokia and the rest of the Symbian ecosystem actually led the market by a long stretch, just a short while ago. If they hadn't hired a former Microsoft exec to lead the company, and perhaps with a bit of luck, Nokia/Siemens/etc would have been that alternative. But that is another discussion.

I 100% disagree, and that's as someone who was both a fan of Nokia and even of Windows Phone.

And even if I agreed, they did hire that former MS exec. So they wouldn't have been that alternative, because in no universe would Apple or Google put Stephen Elop in charge of iPhone/Android, and in this universe, Nokia would.

> The solution is simple: build a business environment that would allow a home-grown alternative to have developed over the last 40 years.

Time travel is a simple and elegant solution to many problems. (Ignoring unintended consequences at least.)

True, but we always have the opportunity to eliminate a similar need for time travel 40 years from now.

I agree, and I'm not sure whether the reason it hasn't happened is that they can't do it, or they won't do it.

These are not mutually exclusive.

I think there could be a big market for a hosting+support provider that manages the patchwork of open source business applications. Once that's set up, the organization could spend money on the development of the systems they're hosting.

I'm thinking a portfolio of auth, storage, chat, email, code repository, project management... Everything an organization could in theory host itself but realistically does not have the personnel for.

I'm halfway there with Communick. I started with the focus on providing hosting for social media and messaging platforms, so I had to find my way around setting up LDAP for SSO, provisioning of object storage for separate services, etc.

But the most interest thing is that in the process I also wanted to remove my dependency on the other centralized SaaS, so I ended up setting up my own git repository (gitea), my own CI (woodpecker), my own project management tool (Taiga), my own knowledge-base/data sharing tool (Baserow).

On the one hand, I agree with you and think it could be a great business opportunity. On the other, the whole thing is so easy to be completely commoditized that I don't see a practical path to profitability. If I go to investors with the idea, they will say (rightly so) that there is no easy way to establish a competitive advantage. If I bootstrap (like I have been doing with Communick) I can not be fast enough to do both customer acquisitation and development.

Isn't that essentially just any existing systems integrator? There's plenty of those that are non-American.

That aside, governments have the resources to do this themselves, that's how they currently do so. Extending those services to local organisations would be a step in the right direction.

The thing about services and tooling is that for many orgs, there's not a whole lot you need, and once you're at a scale where you need tooling to manage that scale, you presumably have the resources for an operations team to deal with that, and can outsource the bits you can't do.

The org I work for outsources our public-facing website work to a web-design co, because that's their speciality, and not ours.

All of that is to say; I agree with you, but I think they already exist in the form of SIs.

> What is important that there is one company you can go to who does all of that for you.

This is what gets us in this mess in the first place.

> Running a government institution and having 20 different vendors to make your basic IT system work is a nightmare.

Then scale down the bureaucracy and bring back the decision-making power down to the leaf nodes. Have each institution working as a "microservice" which is responsible only for defining the interfaces on how to interact with them, but leave the internal implementation completely up to the department. You can of course have some collaborative structure where these departments can use as a reference guide, but they are completely free to override those decisions when it best suits them.

God, the idea of microservices for humans is nightmare time. I work for a company that runs micorservices, and I can say I've spent days attempting to get everything running on my dev system. One upgrade and I can watch my whole day/week disappear into config hell.

I can't imagine how hard it would be to do this with people. Each working with their own little bubble.

Just the other week someone decided that an api needed a tweak, so they adjusted the code and the tests, but missed one external system. Took 4 days to fix, because we couldn't figure out what had changed. And the team who owned the external system wasn't around. People as microservices.. no just no.

> the idea of microservices for humans is nightmare time.

Works for supermarkets, department stores etc. Companies employ too much red tape in their acquisition processes.

I’ve seen organisations pay way over the going rate for cloud services by insisting on a bidding process and talking to salespeople, when they could have just purchased direct from the console.

If doing your own work requires you to "get everything running on your dev system", are you really working on a service-oriented architecture or was it that your company decided to board the bandwagon and botched the execution?

> people as microservices

No, departments as microservices.

> Then scale down the bureaucracy and bring back the decision-making power down to the leaf nodes. Have each institution working as a "microservice" which is responsible only for defining the interfaces on how to interact with them, but leave the internal implementation completely up to the department. You can of course have some collaborative structure where these departments can use as a reference guide, but they are completely free to override those decisions when it best suits them.

Dear god no. That's how you end up with contracts assigned to "Joe's Nephew Software Design" that don't just smell but reek of nepotism (although I will admit, the "big bodyshops" aka Accenture and friends aren't much better), neverending GDPR et al. compliance issues, and massive employee overhead in training and onboarding costs when every local government does its own shit and economies of scale can't be leveraged.

Also, even assuming "Joe's Nephew Software Design" manages to complete the DMV software on time and in budget... who's guaranteeing that in 10 or 20 years Joe's Nephew will still be around to provide updates? It's (way) easier and cheaper to do continuous maintenance when there are lots of clients to fund upkeep, compared to just one.

> the "big bodyshops" aka Accenture and friends aren't much better.

You said it yourself. Corruption and abuse of power will always exist. But if I had to choose between the invisible corruption of faceless bureaucrats enabling cronies or the local crook who will try to put his finger on the pie, I will take the local crook every time. At the very least, I can get a bunch of people and bang on their doors with some pitchforks.

> who's guaranteeing that in 10 or 20 years Joe's Nephew will still be around to provide updates?

We are taking about a scenario where open source is the norm and the stakes for each individual project are lower. "off-the-shelf" components would be the norm. Whatever customization or improvements done by the departments would also be released as FOSS.

> At the very least, I can get a bunch of people and bang on their doors with some pitchforks.

For that, you gotta hear about the issue first. Local reporting is all but dead, and the few local journalists that remain and have the expertise and time to do investigative pieces on local money wastes are way more easily silenced by SLAPP lawsuits and political pressure (up to and including death threats) than something like, say, the New York Times.

But if we are talking about local services and the governance of projects at the municipal/county level, you won't need to wait for reporters. You will quickly see and experience the mismanagement of resources.

I work for a government institution and I assure you that we have more than 20 vendors for IT.

Of course. But your basic IT system, presumably, is a Microsoft system. On top of that you are deploying many more systems, for all the kinds of different use cases.

If you replaced that Microsoft system right now you would have to find individual vendors for each of the parts that Microsoft provides. Getting them together would be a huge nightmare, because even the basics do not work.

This doesn’t seem right. What is Microsoft supplying? Windows which is used almost exclusively to access some web service CRUD form. All of these services are made by third party vendors. Any number of OSes could do that from linux based or ChromeOS or MacOS… probably even iOS. There are some legacy win desktop apps that are slowly getting replaced or they are run in VMs.

The Microsoft servers are most likely azure running linux. Thats quite possible to replace by any number of vendors.

The main MOAT microsoft has are the contacts and the lobby. There always is some politician around fighting for Microsoft because they like Outlook more than Thunderbird.

It’s also reason why i think they will keep their dominant position. Even though the idea they provide something rare is increasingly more untrue.

Not really.

The end user devices are Windows 11, we use M365, but government services are mostly homegrown and the infrastructure runs on Broadcom (VMWare) and IBM (Openshift) software.

Replacing Windows 11 with some kind of Linux and M365 by an MTA is technically feasible, there is political momentum building against US-centric services, but here in Europe politicians are historically highly suspicious of technicians, so nothing gets done yet.

It's a rich country, COTS replaced a lot of technical excellence, but the trend can be reversed as we have bright engineers on the inside still.

In poorer countries and regions, the engineering excellence is way better and they are much more independent.

I'm also working on a government institute, but unlike you, we're absolutely owned by Microsoft. Disruptions to that relationship could be existential threats, which is why (slow) movement has started on detaching from them.

Yep years ago I was a palm-os dev and worked for cities. One government offical told us to switch to windows mobile, because "no one ever got fired buying microsoft"

RedHat and SuSe do compete there.

> Running a government institution and having 20 different vendors to make your basic IT system work is a nightmare.

Let's suppose that is true, because it is. But how is that different from any other entreprise, commercial or public?

There is a difference between having 20 and 40 vendors, though?

Yes. If you want your vendors to interop with each other in any way, it's the same n^2 lines of communication problem you have in dev teams. In fact, it's worse, because the vendors are antagonistic towards each other - it's in their interests that you ditch some of the others and give more of your business to them.

That depends on the vendor. Small vendors know that they can't do everything and are happy they are part of the pie. Medium sized often dream of getting big and so if they think they can by taking a large slice they will.

It also depends on how your relationship is structured and what you demand. I work for a very large company, but some of our customers won't even look at us until we pass a third party interoperability certification, and thus getting that certification becomes critical to us even though most customers don't care. Once we are certified interoperability issues are rare (they happen all the time because of the sear number of customers, but most of the time things just work because everyone is following the standard). The standard and certification has been refined over a couple decades now and so most of the things that can go wrong either are either updated in the standard and certification test; or they are at least tribal knowledge of "don't do that it won't work"

Sometimes one large corporation feels like working with multiple smaller vendors. Products are too siloed and don't work together the way they claim, etc.

Add integration between all the parts to it and you will see why those big companies stay successful.

Not only is managing 20 vendors a nightmare, they all live in their own bubble and moving data from one to the other is normally not that easy.

Using standards typically makes a big difference. And having redundancy, so that lack of interoperability/lock in is actually not something you find out after it is too late.

No organisation of any size buys everything from one vendor though. Microsoft dominates desktops, but Apple and Google dominate mobile devices, an organisation might have Oracle databases running on Linux servers on top of that, some SaaS suppliers, some desktop software suppliers....

But still - think how much more you'd need to buy and validate it all works together. Microsoft gives you AD, which works with Outlook, Sharepoint, Azure, Office365, Teams, then all of those, plus Excel, Word, Powerpoint all bundled, and not for very much money.

True, but it varies depending on how well those fit your organisation's needs whether buying into the full bundle is what you want to do. The less of it you want, the less the advantage.

Most people using MS desktops use AWS rather than Azure. Lots of software from other vendors does reliably work on Windows.

Microsoft has a terrible history of integration even among it's own products and has forced obsoletion throughout. If it's literally you only have a single vendor to pay then you must look for a nationalised solution otherwise you'll just be creating oligarchy.

You can't on the one hand maintain the myth that there will somehow be private competition but then on the other set the barriers so high that only the largest most entrenched monopolies can succeed.

> you must look for a nationalised solution otherwise you'll just be creating oligarchy

Er, why? If France buys a lot of Microsoft licences, they are suddenly an oligarchy?

Reply on news.ycombinator.com