Fog Creek Software
Discussion Board

Is Govt use going to kickstart Linux?

The recent Munich-Linux thread sent me away thinking.

If Linux's biggest problem is the chicken-and-egg issue (commercial apps not being developed because linux isn't on the desktop, and linux isn't on the desktop because there are no apps*) then maybe govt spending is the way to break the cycle.

Businesses need to make money, and hence their view (especially of  IT spending) is short-term. Almost all decisions for a company are on a TCO basis, if not directly, then indirectly.

But governments don't have simple "making money" as their primary goal.  In fact govts don't "make money" in the traditional sense.  They're there to _spend_ money. On infrastructure. Like roads, and bridges and so on.

So govt can take a hit now, in the interests of long-term gain?

To be honest I'm not sure if even govt spending will be enough. At the end of the day, business goes back to the TCO, and I'm not yet convinced that "voluntary" software can compete with "commercial" software. Possibly the products might be technically equivalent, but he comercial software always seems to win the marketting game.

* before I get a  zillion references to all the linux apps out there - I'm speaking generally here.  Of course there are linux apps, and tere are linux developers, but the distribution with windows apps and developers is unequal.

Bruce Johnson
Friday, July 18, 2003

why do people keep saying there are not enough good desktop app's for linux?

really, i think people are heavily underestimating this

Guyon Morée
Friday, July 18, 2003

I think that what we need is enough government expenditure to keep a basic platform. So that a private company will need to provide value-added cum agility, which is what they are  supposed to be good at.

Stephen Jones
Friday, July 18, 2003

I believe the market has shown more than capable of growing a healty IT sector, so that the and fail to see the need for the software sector to be "nationalized".

Just me (Sir to you)
Friday, July 18, 2003

Hi Guyon,

The problem isn't that there aren't apps out there, but that they're not the right apps.

For example, the main consumer accounting application in South Africa is called Pastel. It's probably got like 80% or more of the market (I'm guessing).  Now the problem for all these users is not that there are no accounting apps, but that Pastel doesn't run on Linux. 

I speculate that the makers of Pastel don't make a linux version, for the same reason they don't make a mac version. It's simply economics. Why devote resources to 5% of the market when you can devote them to 90% of the market?

ie Pastel doesn't make a linux version because no-one is asking for it. No-one is asking for it because no-one has linux (* same caveat as first post applies).

The same problem actually applies with software regardless of the commercial size. In Munich they'll have just as big a problem converting the 178 custom apps as they will have converting all their documents from Word & Excel. I'm not saying the problem is insurmountable, just that it's more pain than it's worth for business.

For example I could change to linux. All I have to do is a very detailed software audit to see what I'm running now. Then I have to find linux equivalents. Then I have to retrain everyone (and I mean everyone) on all new software. From the mail client, to the accouting, to office apps, to all the custom stuff.  The pain is not inconsiderable.  Sooner or later a director asks "Why are we doing this?"

Of course there are ways to reduce the pain.

Where possible, today, start converting to apps that
support both windows and linux. For example swapping to OpenOffice or StarOffice can be an interesting experiment. (I tried with StarOffice 5.2 but interoperability with others was a problem.) Start with 1 department.  Ditto the mail client and so on.

What happens though (at the moment) is that sooner or later you hit an app that won't budge. Maybe the accounts folk don't want a new package. Maybe you can't afford to lose all the old data. (and it's not just about proprietary file formats - having an XML export of your old accounting system is helpful, but not sufficient...).

Ideally the day you change the OS should be a non-event, folks are already accostomed to the apps they are already using.

question is - can govt provide the impetous that would encourage software developers to invest the massive resouces it would require to create, and maintain, linux ports.


Bruce Johnson
Friday, July 18, 2003

Hi Sir,

Yes the market has been great at growing the IT sector. I'm not convinced that the market will be as good at changing the sector. 

Commercial software development did move away from the multi-platform days of the early 80's for a reason.  And it's going to take an equally compelling reason for a move back.

Software developers don't develop for linux because the windows market is bigger for the same (or less) effort. They don't develop for linux because it's better business to be great in 1 space than average in 2.  And to be honest, most developers simply don't have the resources to code for both.  It's a chicken, and an egg.

BTW I'm not suggesting Nationalisation - not at all. What I'm saying is that govt can see that diversification, and choice, is good, and hence take a longer-term spending point of view.

If the market was split 50/50 between windows and linux then the real winners are developers. We get to make everything twice, and fundamentally develop twice as much software.


Bruce Johnson
Friday, July 18, 2003

Why does Linux *have* to be a market success? The market has Windows, and that seems to be satisfying most people just fine.

Is there a need for Linux somewhere?


Friday, July 18, 2003

I would say it's most appropriate for local and state govts to invest in free software.  Definitely not the US federal govt, or most other fed govts that work closely with the US. 

Munich was lucky because Germany standardizes on PDF instead of MSWord .doc format, though I'm sure .doc is used a lot for informal communication.

Friday, July 18, 2003

Ýes, Philo. In Psychiatrists offices so that they can prove to all those ripped-their-hair-out-slammed-head-into-keyboard-put-keyboard-through-monitor-(head-still-attached)-windows-programmers that things don't have to be quite so frustrating.

Visiting soon
Friday, July 18, 2003

Hi Philo,

Careful - you'll start something with statements like that <g>...

I should have been a bit tighter with that premise. 

Personally I'm not sure that having 2 "main" operating systems is a good thing.  As a developer myself it would mean either (a) more costs to support 2 platforms or (b) reduced sales from a reduced market.  Either way it translates directly into more expensive software.

On the other hand I do understand the argument that in the current situation there's too much power in the hands of a single company. (I'm not 100% sure I agree with the whole argument, but I do _understand_ it).  I don't think we're necessarily held hostage to MS - but there's no doubt that the existence of macs & linux help to keep redmond from getting carried away.

Ideally we want the cost benefits of competition on our supplier, but with the development overheads of a monopoly.  Um - in other words the status quo. :-)

[Aside : Perhaps what windows developers should do is post lots of notes, and references about changing to linux. Keep the pressure on MS up.... As long as MS is threatened by linux, the big winners are ... um .... MS users... ]


Bruce Johnson
Friday, July 18, 2003

There is not so much a need for Linux, as a need to defend a society against a single point of failure.

A computing platform is 'merely' infrastructure, it is the applications that give people and businesses value.
As with any infrastructure, it should be as guaranteed as possible. Meaning the risk of it disapearing or changing under your feet should be as small as possible.

Traditionally, that is why infrastructures have been subject to government regulation.
Imagine if roads where held privately without any regulation. You would be at the mercy of the road owner. In fact, people where ...

That does not mean that things have to be subject to government regulation to turn out well, but there need to be a common interest.
Electricity companies do not each provide there own voltage level, although they could have, and maybe even did, when they started out.

Given enough room, business can very well recognise the need for standardisation and cooperation at certain levels.

But if no such room exists, for whatever reason, as appears to be the case with Microsoft and desktop computing, something has to happen.

Microsoft does not need to go away, nor does Windows, but there needs to be some sort of protection of application developers and end-users. And it is unlikely that a business can provide such protection without conflict of interest.

More Linux, or Mac, or whatever, could be the start of some move towards standardised computing platforms.
I personally doubt that, but who knows.

Practical Geezer
Friday, July 18, 2003


let us get one thing straight: There is no such thing as "free" software. Someone pays, and that someone is you. Either you pay through taxes, or through marked-up compliments that subsume the cost.

As for the "developing everything twice" thing: I don't know about most of the people here, but when I came into this field, it still had a kind of "future now" aura. I like doing new things. Things that were not possible before. I like automating repetetive things. Reversing things to where I have to do something twice is not my idea of fun.

Just me (Sir to you)
Friday, July 18, 2003

The reason GNU is so pedantic is because they /don't mean low-cost/ when they say "free."  This whole confusion of Free leads to incredible mistakes.

The Germans nailed it when they said Linux was strategically better than Windows.  Maybe they didn't say "GNU/Linux," but they had the spirit.

Now, of course IBM/Suse may start doing their damndest to lock Munich in.  But at least there's one less huge layer of lock-in.  Microsoft may be quite benign compared to companies like Sun, but there are strategic reasons why governments still don't want such a dependency.

Friday, July 18, 2003

"Why does Linux *have* to be a market success? The market has Windows, and that seems to be satisfying most people just fine.

Is there a need for Linux somewhere?"

yep :)

there is a market need for a cheap alternative to the main commercial OS's

TCO aside, if a school can take old, worn out pc's donated by parents, install linux on them and have their children learn how to use computers, how to trouble shoot them, how to browse the internet, how to run an email client etc etc etc then all is right with the world :)

Its not enough to have to rely on charity from big companies, sometimes they deliver (when its worth their while) and sometimes they dont.

As well as schools their are poor families, poor countries, poor hospitals etc etc

...and there is a whole copyright area which makes me eternally grateful to linux, I make my money selling software and that is good, but there is a real need for public domain software that anyone can use, look at the source of, learn from etc etc etc..

Linux is _good_, in every sense it will improve our world and long term I genuinely believe it will make more information available to more people  than microsoft, apple or any other commercial operating system will manage.

It is potentially a threat to all of our incomes (over time there will verra likely be decent OSS versions of pretty much everything), but that will not happen for a loong time if ever, and what the world will receive in exchange is far more valluable in the long run.

If microsoft, apple and ourselves can survive side by side with OSS then good, but if not I would still rather have OSS available to all who need it.

Friday, July 18, 2003

"There is not so much a need for Linux, as a need to defend a society against a single point of failure."

This whole thread reeks of over-self-importance in the world. Instead of wasting time here, you should be working on a replacement for non-renewable fossil fuels. That would have an ACTUAL impact on society.

Seriously, every piece of software you have -- from the OS to the accounting package -- is a "single point of failure". Just because there is a competing OS, and a competing accounting package, doesn't mean you're any less locked in. EVEN if the #s were 50/50 for two different software packages, you'd eventually have to choose one, and then be stuck.

Brad Wilson (
Friday, July 18, 2003

"That would have an ACTUAL impact on society."

I agree, it would :)  but I totally disagree with the implication that Linux/OSS _wont_ also have an actual impact on society.

Would you deny that microsoft has had a huge impact on society? 

Friday, July 18, 2003

" Just because there is a competing OS, and a competing accounting package, doesn't mean you're any less locked in."

That is correct, right now. But the point is that that is exactly what you want to avoid.
Why is it that you are locked in? I assume you mean because you have time invested learning to use the application, and because your data is stuck in it, right?

As long as that does not change, it does not matter much whether you could change OS, or application.

But it does not have to be that way. Business that build applications are in fact already moving to standardised storage formats. It will be many years before you can move between applications with the same data store, but where it used to be very difficult to share data, even if the company that built your application was kind enough to publish the proprietory format, it is now becoming much more feasible already.

Once your data is less locked, you get freedom to select applications based on current merits, not on a historic choice. Obviously you do not do that on a whim, but now you can hardly do it at all.

It will be even longer, if ever, that computing platforms become standardised to the extent that applications built on one platform, will also work on another, without unreasonable hassle.
It will take much more that standardisations on APIs.
But if it ever does, you can select a vendor based on added value, instead of necessity.

Compare the different Linux distributions. This is just my guess, but I assume the difference between them is based on added value, and that it does not matter which one you choose, your applications will work on any of them, assuming they don't rely on any of the features that differentiate them of course.

Will it ever happen, who knows.
Should it happen? I think so.
I am always glad I can buy which ever DVD-player/washing machine/car/... I want and only have to compare on price, or quality or additional features, or whichever criteria I choose.

Practical Geezer
Friday, July 18, 2003

If there were not a need in some fashion Linux would not have made it out of the "interesting toy" stage, but it has. 

Philo -  Is there a need for more than one car maker?  More than one brand or supermarket or more than one maker of Orange juice?  In Linux's case, it had the disadvantage that the market, MS, controlled the future, so it is in catch-up mode.  But managed anyway.

As for applications, I believe we are seeing a natural progression.  The first ones to convert were server side.  Databases, web servers, etc.  These all had near instant ROI and made them easy choices.  As each one was added, more people became aware of, and understood Linux.  As more do, the next step, more desktops and more applications will follow. 

On the idea that developing applications for two OS platforms is "twice as hard", it is only so if you locked yourself into features of the OS that are unique and those components were not isolated into a specific layer.  At the time, this may have not been necessary, because what did you get for it other than design?  Today, everyone should be looking to layer because if it is not Linux, it will be something else. 

Whether that is worth it for your specific market, will only be determined if you can enter it.  If the cost of entry, an entire rewrite of your application, is too high, then the competition will make a competing product and they may very well set it up to run on multiple platforms, making you extinct.  If you think that the OS should prevent that or that Linux or any other OS/platform item should prevent that, then you are placing a bad bet.

Mike Gamerland
Friday, July 18, 2003

We've had a decade where business managers thought software development was easy, and consequently made millions of mistakes in project commissioning and strategic direction.

I think this is just the decade where government will catch up and go through the same process.

Friday, July 18, 2003

"Why is it that you are locked in? I assume you mean because you have time invested learning to use the application, and because your data is stuck in it, right?"

Every software company I've ever worked for has set as one goal to lock customers into our products. Frankly, I'd think it was bad business practice to do otherwise.

Brad Wilson (
Friday, July 18, 2003

Brad, two quick points.

Every company wants lock-in, but no organization wants to be locked in.  The most successful companies, like Walmart and Microsoft, actively avoid dependencies.

Also, the best thing a programmer can do to find alternatives to fossil fuels, is by making inexpensive websites using Free Software for those projects.  Journalists don't usually find things like biodiesel sexy enough to write articles about.  So the biggest cost is sunk trying to inform people.  Programmers shouldn't try to play scientist, but instead teach how putting an ad on Google is a cost-conscious way to go.

Friday, July 18, 2003

Guyon Morée, that's because most of the people talking about oh how Linux is not suitable for this or for that, actually, don't have experience with Linux. Remeber this is JoS and Joel is ex-microsoftie and writes for windows platform and most of the people here write for windows and use windows.

Friday, July 18, 2003

All this thread reminds me how about 5 years ago network admins prefered BSDs. They used to say how superior the implementation of network protocolswere in BSD and how Linux couldn't be used for network applications. Well where are BSDs now? Linux has come a long way and is the best platform for network solutions right now, be it a router or a firewall. The same thing could happen with applications, it just that the things like that don't happen overnight. There is a critical mass of people willing to write for Linux. And even though the first crop of business applications is not perfect, they will be improved over time. Look at Microsoft, they defenitely feel the heat.

Friday, July 18, 2003

If you change from windows to Linux you have to change everything.  This overwelms the users because instead of doing their work they have to spend ton's of time learning the new system and applications.  Their productivity will be shot to 0 until they can learn the new system.

I think it makes more sense slowly if you want to move to open source.  Start with open source applications (OSA) that run on win32 and open source.  Start with IM, email, and maybe even OpenOffice.  Add it slowly so user can adjust over time.  Considuring that licesnces for Exchange and MS Office cost more then Win2000 you could save a lot of cash.

If i was the CIO i would start standerdising on the Mozzila web Browser.  Joel switch to it because it works like IE and blocks pop up adds.  Just think all users hate popup ads so your first step towords open source will be a sucsess.  You get good word of mouth, which is the best advertising.

Friday, July 18, 2003

It's not just Windows vs OSS; the level of expertise is lower among OSS programmers, and that's significant because the types of problems that need to be solved are getting harder.

Friday, July 18, 2003

Now that's just ridiculous.

Friday, July 18, 2003

"It's not just Windows vs OSS; the level of expertise is lower among OSS programmers"

Do you really think that it takes more expertise to write yet another pretty interface for data in a database (what most of the people here probably do) than to write an operating system or RDBMS?  Try following the development lists for either FreeBSD or PostgreSQL for a while and then get back to us.

It's quite amusing to watch a bunch of ignorant Windows developers discuss their "enemy", which they clearly don't understand at all.

Friday, July 18, 2003

"It's quite amusing to watch a bunch of ignorant Windows developers discuss their "enemy", which they clearly don't understand at all."

speak for yourself :)  Im a windows developer primarily and I have a very good understanding of Linux and OSS.

This is _not_ a windows/oss thing, it is a commercial vs OSS thing.

There are many developers who feel threatened by OSS, and who can blame them :)

Its my opinion that both are necessary and good.

Friday, July 18, 2003

As I've said often - I've done the linux thing; dumped it because I got tired of linux and the attitude of the "linux community"
OTOH, I pushed CVS for our source control and I'm learning NANT and Draco.

I have a deep respect for a lot of the OSS coders out there. I just wish they'd channel some of their MS hatred into their work and make it even *better*. [grin]


Friday, July 18, 2003

Actually I have followed some of the OSS lists from time to time. That's part of what I base my views on. You think that stuff's advanced, huh?

> Try following the development lists for either FreeBSD or PostgreSQL for a while and then get back to us.

Friday, July 18, 2003

*  Recent Topics

*  Fog Creek Home