[Info-vax] An alternative history of computing

[John Wallace]

On 28/07/2021 17:26, Bill Gunshannon wrote:
> On 7/28/21 9:41 AM, Dave Froble wrote:
>> On 7/28/2021 5:45 AM, chris wrote:
>>> On 07/28/21 02:50, David Jones wrote:
>>>> On Monday, July 26, 2021 at 6:27:22 AM UTC-4, chris wrote:
>>>>> Originally, all that was loosely based on an ISO model,
>>>>> the sort of standards that DEC were great supporters
>>>>> and contributors to at all levels, but really backed
>>>>> themselves into a corner over decnet. An obscure set
>>>>> of protocols and command set reminiscent of the sort of
>>>>> serisl line and telco ideas dating back to the 1970's.
>>>>> TCP/IP was faster, easier to visualise in design, to
>>>>> program and above all, a completely open source and fixed
>>>>> set of standards that anyone could use, improve and generally
>>>>> contribute to.
>>>>
>>>> The TCP/IP standards were developed over decades, RFC superceding RFC.
>>>> The RFC specifications often had gaps which results in conflicting
>>>> implementations
>>>> by different parties, usually resolved by adopting the interpretation
>>>> of the one
>>>> which has bigger presence in the rather limited ARPANET ecosystem.
>>>>
>>>>   Eventually it got reliable enough that  V.P. Gore proposed dropping
>>>> the commerce
>>>> restrictions.
>>>
>>> That's a rather biased, one sided view. I was using tcp/ip in the late
>>> 1980's
>>
>> Well, there is your problem.  In the late 1970s DECnet was a working 
>> product. 
> 
> A working product that only worked on a very small subset of computers
> in use at the time and with little or no long distance capabilities.
> 
>>           As I remember, not so for TCP/IP.  I'd perhaps suggest that 
>> you need to be a bit older, but, it's no fun, won't wish that on someone.
> 
> One really has to wonder why, if it had a decade head start OSI (aka
> DECnet) never acquired the capabilities of TCP/IP.  Why is the Internet
> TCP/IP based and not OSI?
> 
> bill
> 
> 

The internet is still largely IPv4 based, even though IPv6 has allegedly 
been around almost as long as OSI implementations.

Retail ISPs back then liked cheap stuff and cheap staff. Mostly they 
still do. IPv4 was cheap. Browsers were cheap. Expertise and training? 
Hmmm, no, we don't do that, it costs too much.

IPv4 is still cheap (for ISPs). IPv4's ongoing impact on end users 
varies, but apparently things like CGNAT (widely seen in the mobile 
world) are not much fun for end users who want to go slightly off 
mainstream.

In the user-visible world, insecure browsers (and insecure email, and 
...) are now a huge contributor to IT insecurity. But never mind, 
Tikstabook will sort it. Perhaps.

Also at the application level, prior to proprietary non-standards-based 
web-GUI-based email taking over the world, the main setup that the IP 
world could offer was a teletype-era mail setup (POP/SMTP) which needed 
dozens of band-aids to be added before it was actually anything like 
usable, especially in a multiplatform environment. That's before we even 
get into "advanced" higher-layer topics like authentication, 
authorization, and optional extras such as (e.g. for email) secure 
delivery, non-repudiation, etc.

Still I guess some people must see it as progress.