[ih] Fwd: [Dewayne-Net] The Rise and Fall of the Gopher Protocol

[Brian E Carpenter]

On 21/08/2016 14:54, Dave Crocker wrote:
> On 8/20/2016 6:31 PM, Brian E Carpenter wrote:
>> In fact, hypertext
>> specialists thought it was broken; 

The academics working on hypertext thought that a hypertext system
without bidirectional links and some kind of continuous completeness
checking (so that dead links would vanish automatically) wasn't
useful. I remember a public argument between Tim BL and Hermann Maurer*
(of Hyper-G fame) about this - I can't put an exact date on it, but
it was when Hermann gave a seminar  about Hyper-G at CERN and I'm
guessing it was in 1993/94. (Maurer's later writings assert that Hyper-G
was built on WWW experience, but my recollection is that if so, that
can refer only to the very early pre-Mosaic web.)

Tim made a very strong argument that a system with bidirectional links
and consistency checks was undeployable at large scale, and of course
he was right.

*My memory says it was Maurer. Some Google hints suggest that it was his
student Frank Kappe.

All the same, Maurer wrote in late 1994:
'So is WWW the answer we all have been waiting for? Unfortunately, the answer is
again a clear: "NO".'
(elib.zib.de/pub/Workshops/TU_Berlin_1995/Maurer/Maurer.ps)

Apparently, the Microcosm people at the University of Southampton had similar
criticisms of the WWW design.

>> distributed systems designers thought it
>> was broken. 

In particular, they thought POST was broken because it didn't offer transactional
integrity. And they still do, I think. Google "RESTful considered harmful."
Or think about how disasters like XML-RPC and SOAP arose.

    Brian

>> I suppose gopher was the same. This flatness actually made deployment
>> a great deal easier.
> 
> 
> No doubt I wasn't tracking any of this closely enough, but I don't 
> recall hearing those complaints.
> 
> But then, my framework for such things was thoroughly imprinted by 
> having gotten access and becoming a longtime user of the Engelbart NLS 
> system, starting in 1972.  (My start; the system itself dated back to 
> the 60s.)
> 
> It had the same, at-will, direct, inter-document linking (albeit not 
> inter-machine).  Any place in any document could include a link to any 
> other labeled/numbered place in any other document.