Re: No call for Ada (was Re: Announcing new scripting/prototyping language)

[Dmitry A. Kazakov <mailbox@dmitry-kazakov.de>]

On Tue, 27 Apr 2004 07:26:01 +0400 (MSD), "Alexander E. Kopilovich"
<aek@VB1162.spb.edu> wrote:

>Dmitry A. Kazakov wrote:
>
>> >Modern science deals also with budgets, management, conferences, grants,
>> >degrees and citations.
>>
>> that's scientific bureaucracy 
>
>Call it what you wish, but take into account that vast majority of scientists
>do not oppose it anymore, but participate in all that stuff rather actively.

As cynical as "people of Ethiopia do not oppose hunger, but actively
participate it".

>> >By the way, just curious, did you sell *anything* *on open market* (shareware
>> >or commercial product)?
>>
>> The firm I am working in, sells several commercial products.
>
>There is a huge and principal difference between selling your products on open
>market and working for a firm that do that. When you are an individual seller
>you take decisions and see immediate results.

Where you saw individual software sellers? The products we are
developing have weight of many man-years.

>You choose the price for your
>product, and you income depends on whether your guess is right. But when you
>work for a firm (not as a top manager, but as a software developer), you do
>not take such decisions, and your salary does not immediately depend on the
>particular results. Yes, it is possible that you may somehow influence your
>firm's decisions regarding product prices and marketing policies in general,
>but this is very far from taking decision yourself. If you try that once then
>you'll see this acute difference.

What's your point?

>> Though it
>> is not clear what sort of experience you meant,
>
>I meant practical experience - just create at least one software product and
>sell it yourself - either as a invidividual or as a top manager of a firm,
>regardless of the firm's size. After that you most probably will not be so
>sure that liability for software products is a good thing.

Strangely enough that they are sure for almost all other things...

>Recall quantum mechanics - it is a science, and there are things, which it
>can predict, but there are also severe limitations to its prediction power.

You are comparing limitations of something with nothing.

>> I meant only the law, which should treat sold software as an insurance
>> contract rather than a "right to use". At least contract parties
>> should have equal rights.
>
>I don't quite understand that: do you mean that some insurance company must
>be always involved as a mandatory third party between a vendor and a user?
>That is, every software product to be legally sold must be insured by some
>insurance firm? Then who will be really (and legally) insured - vendor or 
>user?

No, a vendor has to be liable to the software product it sells.
Because software products are not "consumed" as normal products are,
there seems to be only choice between "right of use" and "insurance"
models. "Right of use" model is what we have now. Its disadvantages
are quite clear:

1. It works against quality products;
2. It effectively stops any significant progress by suppressing
competition;
3. It gives customers no protection from fraud;
4. It imposes real threat to basic human rights (see DCMA)

>Note that market never chooses immediately and/or forever. It is a big and
>substantially stochasic system, and moreover, it is a multi-dimensional
>system. Sometimes we can see some clear and relatively stable preference and
>we call it a choice (done by the market).

It is about game rules, which state should impose.

>I can't see what can it mean really
>that "software market chooses the worst".

Sorry for that.

>> >> The navigation system, radio, CD
>> >> player etc, all that will be connected to the field bus. They will
>> >> also have Bluetooth and Internet connections. This will open wide
>> >> possibilities for attacks of all sorts.
>> >
>> >All that is quite obvious, and car vendors personnel (including their
>> >software engineers) are neither preschool kids nor full idiots. After all,
>> >probably they all have cars.
>>
>> There is a overwhelming force of market overriding everything you have
>> mentioned.
>
>How do you know the direction of this force, and whether that direction will
>be stable in near future? You just said that there is no science that can
>predict market economy.

Exactly, this is why your talks about good-will of managers,
engineers, programmers are absolutely irrelevant.

>How do you think, if some car vendors will produce fully computerized cars
>that will be obviously unreliable, and at the same time some other car
>vendors decided to wait and continue production of not-so-computerized cars,
>what will the market choose?

Market already chose. There will be no not-computerized cars. Further
the question is not which features of cars market will choose. It is
about how reliable cars will be. And also, whether our society is
ready to tolerate unreliable cars. So far it was not. Now the real New
Order (=software mess) starts to threat our live in areas we
accustomed to see unconnected to software. This or that way we will
react.

>> And, well, who will serve 25 years in jail for cracking
>> proprietary protocols?
>
>There is absolutely no need to crack anything. First, there probably will not
>be anything interesting enough to crack. Second, all those protocols can be
>easily imagined - well, they certainly will not be the same as in real cars,
>but so what? Those parties who will think that the difference is significant
>for them will point on it and provide an information, which will be sufficient
>for adjustment or generalization. We simply should not care about the distant
>sources of that information, it will be sufficient that information is 
>provided publicly, and it need not be exact information - it may be just a
>hint.

You are mistaken. (I am working with some of these protocols)

>Although certainly it would be better if the sources of actual car software
>were published, as I said some time ago.
>
>> >And that Ada 2005 would be very appropriate programming
>> >language for this purpose.
>>
>> Ada 95 + SPARK (for some components) would be ideal.
>
>Yes, I think that SPARK may be quite useful for some components.
>
>> But I doubt it will be even considered.
>
>Why not? We (or someone else) can consider it. Perhaps you mean car vendors
>here, but why should we bother ourselves with *their* problems?

Because they decide.

--
Regards,
Dmitry Kazakov
www.dmitry-kazakov.de