Software as a Service

[Flyback]

Software as a Service is not a new idea.

I've thought about it in the past but never realized its full potential. The key benefit is simply that you can control access to the product and cut it off at any time if people don't pay. Also, you can go to a subscription based model instead. Less piracy, too. It lets you increase profit.

But something I realized more recently is that in some cases (not all), you could protect something pretty advanced by never actually having to distribute a binary/the technology. There would be no potential to reverse engineer the underlying idea or algorithm.

It would only work in certain types of systems, but things that are heavily algorithm-oriented which rely on some key algorithm no one else has could benefit from this. There would be additional overhead costs like if you need the hardware to host servers for each client to do xyz task (as with all Software as a Service), but you could pass the charge back to them. This would allow you to keep algorithms completely private and undisclosed. If it was something really quite groundbreaking you wouldn't patent it at all (or publish the inner-workings in any means).

Without patenting it or publishing any material, no one can just hop onto the idea when a normal patent would expire (don't you have to detail how it works in your patent application/file ?).

The only idea I can think of where this would work really well is in software that is heavily dependent on algorithms. If you developed some really unique algorithm ahead of its time that wouldn't likely be solved or discovered by anyone else in the near future you could maximize the return on it.

What do you think?

 

[sao123]

Originally posted by: Flyback
Software as a Service is not a new idea.

I've thought about it in the past but never realized its full potential. The key benefit is simply that you can control access to the product and cut it off at any time if people don't pay. Also, you can go to a subscription based model instead. Less piracy, too. It lets you increase profit.

But something I realized more recently is that in some cases (not all), you could protect something pretty advanced by never actually having to distribute a binary/the technology. There would be no potential to reverse engineer the underlying idea or algorithm.

It would only work in certain types of systems, but things that are heavily algorithm-oriented which rely on some key algorithm no one else has could benefit from this. There would be additional overhead costs like if you need the hardware to host servers for each client to do xyz task (as with all Software as a Service), but you could pass the charge back to them. This would allow you to keep algorithms completely private and undisclosed. If it was something really quite groundbreaking you wouldn't patent it at all (or publish the inner-workings in any means).

Without patenting it or publishing any material, no one can just hop onto the idea when a normal patent would expire (don't you have to detail how it works in your patent application/file ?).

The only idea I can think of where this would work really well is in software that is heavily dependent on algorithms. If you developed some really unique algorithm ahead of its time that wouldn't likely be solved or discovered by anyone else in the near future you could maximize the return on it.

What do you think?

software as a service will ultimately fail because people had delusional paranoidness...

CableTV works because it is merely full spectrum content delivery... there is basically no talkback, which channel you watch is virtually undetectable, so there is no "big brother" in your technology...

With interactive programming such as software, you lose the benefit of this privacy...


additionally, the internet does not have the bandwidth to host all applications for all the PC users in the world. Not even close.

 

[Flyback]

Originally posted by: sao123

Originally posted by: Flyback
Software as a Service is not a new idea.

I've thought about it in the past but never realized its full potential. The key benefit is simply that you can control access to the product and cut it off at any time if people don't pay. Also, you can go to a subscription based model instead. Less piracy, too. It lets you increase profit.

But something I realized more recently is that in some cases (not all), you could protect something pretty advanced by never actually having to distribute a binary/the technology. There would be no potential to reverse engineer the underlying idea or algorithm.

It would only work in certain types of systems, but things that are heavily algorithm-oriented which rely on some key algorithm no one else has could benefit from this. There would be additional overhead costs like if you need the hardware to host servers for each client to do xyz task (as with all Software as a Service), but you could pass the charge back to them. This would allow you to keep algorithms completely private and undisclosed. If it was something really quite groundbreaking you wouldn't patent it at all (or publish the inner-workings in any means).

Without patenting it or publishing any material, no one can just hop onto the idea when a normal patent would expire (don't you have to detail how it works in your patent application/file ?).

The only idea I can think of where this would work really well is in software that is heavily dependent on algorithms. If you developed some really unique algorithm ahead of its time that wouldn't likely be solved or discovered by anyone else in the near future you could maximize the return on it.

What do you think?

software as a service will ultimately fail because people had delusional paranoidness...

CableTV works because it is merely full spectrum content delivery... there is basically no talkback, which channel you watch is virtually undetectable, so there is no "big brother" in your technology...

With interactive programming such as software, you lose the benefit of this privacy...


additionally, the internet does not have the bandwidth to host all applications for all the PC users in the world. Not even close.

I wouldn't be so quick to dismiss it. Lots of companies aren't asking--they are dictating--that their newer software comes as a service.

I understand privacy concerns (and that is my largest concern) but it doesn't really matter because a lot of businesses stand to profit from SaaS and will increasingly move towards it.

(I know that SaaS != Web App, but in many cases there is lots of overlap.)
Not all applications can be hosted obviously and things like a "web OS" are just pure bull IMO, but many many apps are moving into web app domain with increasing browser support for a full-on GUI (beyond simple Javascript+AJAX... I mean like XUL interfaces and whatnot).

Again, the idea I had would only work in systems that are really heavily dependent on algorithms and not something like a calendar program.

 

[tweakmm]

Originally posted by: sao123
With interactive programming such as software, you lose the benefit of this privacy...

The average joe cares this much "_" about that.

additionally, the internet does not have the bandwidth to host all applications for all the PC users in the world. Not even close.

Not now, but we will soon. And you're fooling yourself hardcore if you don't think we're moving back to the mainframe/thin client model again.


Hell, I'm developing an ultra niche software as a service program right now that people are going to be banging down my door for.

 

[Descartes]

Everything can be reverse-engineered. Ok, there might be a few exceptions, but people reverse engineer ciphers, so simple algorithms aren't likely to be any problem either.

As you said, software as a service isn't new. It really started when universities (and others) started selling CPU time. Things have sufficiently progressed such that you can sell application stubs and have it load applications on other servers. This was the model early ASPs tried to adopt, but there was little success with that.

My previous company developed several products that we sold as a service. We had Citrix, Unix (mostly SCO) and Windows deployments, and all of which can help with such a concept. The problem ultimately comes down to this: To truly provide a service, one needed to develop something of a kiosk to which you could deploy. IBM had a Thinkstation (I think that's what it was called) that you could use to essentially bootstrap to something like Citrix, and from there you could provide such a service. Unix inherently has this ability, and Microsoft's further adoptance of RDP and other protocols allows you to accomplish something very similar. Java and .NET both provide core abilities to develop software in this model as well. .NET in particular has a very large amount of support behind it, and it's really quite easy to develop these applications.

That said, don't think it can't be reverse-engineered. If you expect the code to execute on a PC, then it can and will be easily reverse-engineered. Runtimes like Java and .NET make this especially easy. If you leave all the execution to the server, then you face the kiosk problem I discussed above.

On another note, you're likely familiar with SOA (service-oriented architecture). It's been discussed in the community for quite a few years (I actually worked with Microsoft on the first reference-architecture for SOA), but it's getting a lot of interest in larger companies. I know of a dozen government agencies that have explicit SOA requirements (a lot of which is derived from their misunderstanding of SOA, but that's another story).

In summary, it's really a lot of recycled ideas. It started before the internet, Windows, etc. and the core philosophies that led to those movements are simply being rebranded in a new deployment and business model.

 

[MikeyIs4Dcats]

the big drawback to software as a service is the multitudes of people like me who for the most part refuse to pay an ongoing fee for software. I want to buy it upfront and own it forever.

 

[tweakmm]

Originally posted by: MikeyIs4Dcats
the big drawback to software as a service is the multitudes of people like me who for the most part refuse to pay an ongoing fee for software. I want to buy it upfront and own it forever.

The problem is going to be when you can't get the software you need as a compiled binary. It's going to happen sooner than later. I remember reading about this years and years ago, it's going to happen.

 

[MikeyIs4Dcats]

Originally posted by: tweakmm

Originally posted by: MikeyIs4Dcats
the big drawback to software as a service is the multitudes of people like me who for the most part refuse to pay an ongoing fee for software. I want to buy it upfront and own it forever.

The problem is going to be when you can't get the software you need as a compiled binary. It's going to happen sooner than later. I remember reading about this years and years ago, it's going to happen.

possibly, but for now, it's not a big issue. People have talked about the possibility of MS doing this with their products for years, but bottom line being, the masses will reject it.

 

[Descartes]

Originally posted by: Flyback

Originally posted by: sao123

Originally posted by: Flyback
Software as a Service is not a new idea.

I've thought about it in the past but never realized its full potential. The key benefit is simply that you can control access to the product and cut it off at any time if people don't pay. Also, you can go to a subscription based model instead. Less piracy, too. It lets you increase profit.

But something I realized more recently is that in some cases (not all), you could protect something pretty advanced by never actually having to distribute a binary/the technology. There would be no potential to reverse engineer the underlying idea or algorithm.

It would only work in certain types of systems, but things that are heavily algorithm-oriented which rely on some key algorithm no one else has could benefit from this. There would be additional overhead costs like if you need the hardware to host servers for each client to do xyz task (as with all Software as a Service), but you could pass the charge back to them. This would allow you to keep algorithms completely private and undisclosed. If it was something really quite groundbreaking you wouldn't patent it at all (or publish the inner-workings in any means).

Without patenting it or publishing any material, no one can just hop onto the idea when a normal patent would expire (don't you have to detail how it works in your patent application/file ?).

The only idea I can think of where this would work really well is in software that is heavily dependent on algorithms. If you developed some really unique algorithm ahead of its time that wouldn't likely be solved or discovered by anyone else in the near future you could maximize the return on it.

What do you think?

software as a service will ultimately fail because people had delusional paranoidness...

CableTV works because it is merely full spectrum content delivery... there is basically no talkback, which channel you watch is virtually undetectable, so there is no "big brother" in your technology...

With interactive programming such as software, you lose the benefit of this privacy...


additionally, the internet does not have the bandwidth to host all applications for all the PC users in the world. Not even close.

I wouldn't be so quick to dismiss it. Lots of companies aren't asking--they are dictating--that their newer software comes as a service.

I understand privacy concerns (and that is my largest concern) but it doesn't really matter because a lot of businesses stand to profit from SaaS and will increasingly move towards it.

(I know that SaaS != Web App, but in many cases there is lots of overlap.)
Not all applications can be hosted obviously and things like a "web OS" are just pure bull IMO, but many many apps are moving into web app domain with increasing browser support for a full-on GUI (beyond simple Javascript+AJAX... I mean like XUL interfaces and whatnot).

Again, the idea I had would only work in systems that are really heavily dependent on algorithms and not something like a calendar program.

I wouldn't be so quick to dismiss it. I was bootstrapping OS' (I had trying to pluralise acronyms that end in S) years ago, and the loader was on a remote server. Loading something as complex as Windows isn't feasible, but again, newer development platforms that load/compile code JIT (just in time) allow you to fragment your deployment model.

 

[Flyback]

Originally posted by: Descartes

Originally posted by: Flyback

Originally posted by: sao123

Originally posted by: Flyback
Software as a Service is not a new idea.

I've thought about it in the past but never realized its full potential. The key benefit is simply that you can control access to the product and cut it off at any time if people don't pay. Also, you can go to a subscription based model instead. Less piracy, too. It lets you increase profit.

But something I realized more recently is that in some cases (not all), you could protect something pretty advanced by never actually having to distribute a binary/the technology. There would be no potential to reverse engineer the underlying idea or algorithm.

It would only work in certain types of systems, but things that are heavily algorithm-oriented which rely on some key algorithm no one else has could benefit from this. There would be additional overhead costs like if you need the hardware to host servers for each client to do xyz task (as with all Software as a Service), but you could pass the charge back to them. This would allow you to keep algorithms completely private and undisclosed. If it was something really quite groundbreaking you wouldn't patent it at all (or publish the inner-workings in any means).

Without patenting it or publishing any material, no one can just hop onto the idea when a normal patent would expire (don't you have to detail how it works in your patent application/file ?).

The only idea I can think of where this would work really well is in software that is heavily dependent on algorithms. If you developed some really unique algorithm ahead of its time that wouldn't likely be solved or discovered by anyone else in the near future you could maximize the return on it.

What do you think?

software as a service will ultimately fail because people had delusional paranoidness...

CableTV works because it is merely full spectrum content delivery... there is basically no talkback, which channel you watch is virtually undetectable, so there is no "big brother" in your technology...

With interactive programming such as software, you lose the benefit of this privacy...


additionally, the internet does not have the bandwidth to host all applications for all the PC users in the world. Not even close.

I wouldn't be so quick to dismiss it. Lots of companies aren't asking--they are dictating--that their newer software comes as a service.

I understand privacy concerns (and that is my largest concern) but it doesn't really matter because a lot of businesses stand to profit from SaaS and will increasingly move towards it.

(I know that SaaS != Web App, but in many cases there is lots of overlap.)
Not all applications can be hosted obviously and things like a "web OS" are just pure bull IMO, but many many apps are moving into web app domain with increasing browser support for a full-on GUI (beyond simple Javascript+AJAX... I mean like XUL interfaces and whatnot).

Again, the idea I had would only work in systems that are really heavily dependent on algorithms and not something like a calendar program.

I wouldn't be so quick to dismiss it. I was bootstrapping OS' (I had trying to pluralise acronyms that end in S) years ago, and the loader was on a remote server. Loading something as complex as Windows isn't feasible, but again, newer development platforms that load/compile code JIT (just in time) allow you to fragment your deployment model.

I'll admit most of my lack of faith in it comes from a misunderstanding. I just think it would take an incredible amount of bandwidth to happen over the coming 5 years. What would a WebOS look like? I think people largely use the wrong terminology--what they have in mind is perhaps something very lightweight. I'll have to read up if I get some extra time.

Further, I still think you could keep some algorithms secret. Everything can be reverse-engineered if you have access. It becomes more difficult with a very complicated algorithm for which you don't have access. Simple ones? Yeah, they wouldn't be so hard.

I guess I just had a particular case in mind when writing the op 😉

 

[Descartes]

Originally posted by: Flyback

Originally posted by: Descartes

Originally posted by: Flyback

Originally posted by: sao123

Originally posted by: Flyback
Software as a Service is not a new idea.

I've thought about it in the past but never realized its full potential. The key benefit is simply that you can control access to the product and cut it off at any time if people don't pay. Also, you can go to a subscription based model instead. Less piracy, too. It lets you increase profit.

But something I realized more recently is that in some cases (not all), you could protect something pretty advanced by never actually having to distribute a binary/the technology. There would be no potential to reverse engineer the underlying idea or algorithm.

It would only work in certain types of systems, but things that are heavily algorithm-oriented which rely on some key algorithm no one else has could benefit from this. There would be additional overhead costs like if you need the hardware to host servers for each client to do xyz task (as with all Software as a Service), but you could pass the charge back to them. This would allow you to keep algorithms completely private and undisclosed. If it was something really quite groundbreaking you wouldn't patent it at all (or publish the inner-workings in any means).

Without patenting it or publishing any material, no one can just hop onto the idea when a normal patent would expire (don't you have to detail how it works in your patent application/file ?).

The only idea I can think of where this would work really well is in software that is heavily dependent on algorithms. If you developed some really unique algorithm ahead of its time that wouldn't likely be solved or discovered by anyone else in the near future you could maximize the return on it.

What do you think?

software as a service will ultimately fail because people had delusional paranoidness...

CableTV works because it is merely full spectrum content delivery... there is basically no talkback, which channel you watch is virtually undetectable, so there is no "big brother" in your technology...

With interactive programming such as software, you lose the benefit of this privacy...


additionally, the internet does not have the bandwidth to host all applications for all the PC users in the world. Not even close.

I wouldn't be so quick to dismiss it. Lots of companies aren't asking--they are dictating--that their newer software comes as a service.

I understand privacy concerns (and that is my largest concern) but it doesn't really matter because a lot of businesses stand to profit from SaaS and will increasingly move towards it.

(I know that SaaS != Web App, but in many cases there is lots of overlap.)
Not all applications can be hosted obviously and things like a "web OS" are just pure bull IMO, but many many apps are moving into web app domain with increasing browser support for a full-on GUI (beyond simple Javascript+AJAX... I mean like XUL interfaces and whatnot).

Again, the idea I had would only work in systems that are really heavily dependent on algorithms and not something like a calendar program.

I wouldn't be so quick to dismiss it. I was bootstrapping OS' (I had trying to pluralise acronyms that end in S) years ago, and the loader was on a remote server. Loading something as complex as Windows isn't feasible, but again, newer development platforms that load/compile code JIT (just in time) allow you to fragment your deployment model.

I'll admit most of my lack of faith in it comes from a misunderstanding. I just think it would take an incredible amount of bandwidth to happen over the coming 5 years. What would a WebOS look like? I think people largely use the wrong terminology--what they have in mind is perhaps something very lightweight. I'll have to read up if I get some extra time.

Web-deployed need not mean web-based. You will always need a loader, because all machines capable of running some non-trivial OS have to look for the initial loader somewhere on disk/memory/whatever. This very basic loader then has the ability to either continue loading from disk, or to instead refer to another server, and that server can be available on any medium. Unix has done this for years, Windows has some ability to do so as well, and even smaller devices (namely Smartphones) use this model to some degree; they give you the initial loader and perhaps a limited user-interface, but the actual applications, etc. are retrieved as needed, cached in memory as needed and discarded when not needed.

I think the key point here is to ignore web anything; instead, just consider the idea of abstracting away the initial loader (in the case of a PC, the BIOS in combination with the system files, MBR, etc. that exist in the first sector of the hard drive) of some device and subsequent operating system it uses. A weak analogy might be to suggest something like MS-DOS being loaded on the machine and Windows 3.11 being loaded when requested. That would be completely feasible for small deployments given current bandwidth availability, albeit at a significant drop in functionality.

Further, I still think you could keep some algorithms secret. Everything can be reverse-engineered if you have access. It becomes more difficult with a very complicated algorithm for which you don't have access. Simple ones? Yeah, they wouldn't be so hard.

My fundamental approach to security is simple: Don't make available anything that you ultimately can't afford to lose. If you're sitting on an algorithm that's generating billions in revenue for your company, someone will find a way to reverse engineer it.

It's a variation of the old security-by-obscurity argument. People think that if something isn't well known, accessible, etc. that it's inherently more secure than that which is more open, but we know the opposite to be true. This doesn't consider intellectual property concerns of course. Consider how some people are turned away from runtimes like Java or .NET simply because they think it's inherently insecure due to the bytecode/intermediate language (for Java and .NET, respectively) used. The reality is that the only different is the level of abstraction, and someone wanting your secrets can just as well do a simple reverse-engineer of your x86-compiled code as they can Java or .NET.

The point: If you allow any of these algorithms to actually execute on a user's machine, it can and will be reverse-engineered if there is thought to be value in doing so by others.

I think I'm repeating myself now 😀

I guess I just had a particular case in mind when writing the op 😉

There is definitely something viable in the model, but the proverbial devil is in the details. How does that saying go? Those who say something can't be done shouldn't stop those that are. So, if you have an idea... go for it 😀

[edit]Thought I'd say I apologize in advance if any of the above is patronizing. Not trying to tell you anything you probably already know...[/edit]

 

[Miramonti]

This is where Microsoft wants to go, which tells us one thing...it won't happen until Google takes the leadership role.

 

[Flyback]

Descartes your posts are quite insightful and I enjoy them very much.

I know how you should never rely on obscurity. Let us say that you had software that, in part, factored numbers. The key selling point was that you developed an algorithm that factored any number in constant time 🙂P).

Now, instead of selling a product whereby they get a compiled binary on their machines that contains the algorithm in it, it farms that code out to your privately hosted servers (and let us say for argument that they are impenetrable). Also, the other component of the system is delivered in a SaaS-like way, so together they never really get anything tangible that -matters-.

If you don't publish the inner-workings of the algorithm nor patent it, and you are the only person who has physical access to the servers, is it still vulnerable to reverse-engineering?

And oh crap--I believe that my idea isn't just SaaS but rather a combination with SOA 😛 I don't know enough about either.

 

[Matthias99]

Without patenting it or publishing any material, no one can just hop onto the idea when a normal patent would expire (don't you have to detail how it works in your patent application/file ?).

The only idea I can think of where this would work really well is in software that is heavily dependent on algorithms. If you developed some really unique algorithm ahead of its time that wouldn't likely be solved or discovered by anyone else in the near future you could maximize the return on it.

What do you think?

Basically, trade secret protection.

Trying to keep algorithms secret is generally only done in cases where you are doing something very time-consuming, and you have developed some new way of doing it much faster than anyone else can. Sometimes it is also done for security reasons (for instance, using black-box encryption hardware), but there are ways to attack such 'secret' algorithms if you have a lot of time and are able to control the inputs and monitor the outputs. As mentioned above -- if it runs on the client's hardware, it's possible to figure out how it works given enough time and effort.

The idea of 'renting' software/services is not new, and I think it's only going to get bigger. Providers like it because you can't pirate services; if you aren't a paying customer, you're cut off. It also centralizes management; you generally just have some sort of thin front-end hosted locally. If you use a common front-end like a web browser, you don't even have to worry about pushing out client updates; you just update the server-side software and the users get the new version the next time they access it.

However, it doesn't work well right now for extremely high-bandwidth applications over the Internet. But you can do things like leasing servers so the application can be run over a LAN, or 'leasing' the customer software that they run on their own servers.

OS bootstrapping (using VMs) is definitely feasable as long as you have the bandwidth for it. Even multi-gigabyte images can be loaded on demand over a fast LAN in tolerable timeframes.

 

[Flyback]

Matthias99 do you know much about trade secret protection? Can you ever be "forced" to hand it over? (if you are ruled a monopoly, perhaps?)