Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!linus!philabs!prls!pyramid!hplabs!hplabsc!kempf
From: kempf@hplabsc.UUCP (Jim Kempf)
Newsgroups: net.unix
Subject: Re: windows on normal terminals
Message-ID: <291@hplabsc.UUCP>
Date: Thu, 12-Jun-86 11:03:55 EDT
Article-I.D.: hplabsc.291
Posted: Thu Jun 12 11:03:55 1986
Date-Received: Sat, 14-Jun-86 03:42:10 EDT
References: <363@hrc63.UUCP> <16600003@ztivax.UUCP> <522@cad.BERKELEY.EDU> <395@dg_rtp.UUCP> <588@cad.BERKELEY.EDU>
Organization: Hewlett Packard Labs, Palo Alto CA
Lines: 53

> >>         In a few years most new terminals will probably be high-
> >>     performance, micros w/ high-resolution bitmap graphics and built-in
> >>     windowing OS software.  Why fool with this obsolete 24-line stuff?
> 						  ^^^^^^^^ ;-) ;-)
> 
> >Second, the terminal that my employer sees fit to provide for
> >me is an "obsolete 24-line" terminal, despite my preference for a $100K
> >workstation.  I can't understand why an $n-hundred terminal and 1/mth of
> >a $100K machine [...]
> 
> The cheapest "reasonable" terminals that I have seen are about $500.
> A Mac running uw (unix windows) costs about $1200 (I think), and
> even workstations don't cost that much more; I think that the posting
> a few months back about cheap *nix boxes concluded that you could by a
> Sun 3/75 with a small hard disk and ethernet connection for about $9K.
> That's about 20X more expensive than the dumb terminal, but includes
> local computing power.
> 

Since the whole point of commercial software is to make money for
the company, one way to optimize the return on investment is to
reduce your per engineer cost. A 20x cost factor for a workstation
v.s. a terminal, spread over a 30 engineer project could very
well be the difference between a commercially viable product and
one in which the development costs kill any chance of profitability.

In addition, no engineer spends 100% of his/her time hacking. In
fact, most studies show that only about 5-10% of a typical product
engineer's time is spent doing coding. The rest is meetings, reports,
etc. etc. During the time no hacking is occuring, that $9K capital
investment is basically idle and losing money for the company. 

There are two solutions to this problem. Either you multiplex
workstations among engineers or you get dumb terminals. Since
multiplexing leads to scheduling conflicts and constraints on
the engineer as to when she/he can hack, the latter is more
attractive. 

As an interesting side note which might reflect on the future
of the personal workstation, several years ago when personal
computers came out, market research firms were predicting 
enormous sales on the basis of one computer per desk top.
The actual sales were only one third of the predicted number,
because most employers were only buying one computer per
three employees. I think that something similar will probably
happen with personal workstations.

Of course, if you're in academia, you don't have to justify 
your capital equipment costs, except to the NSF, and they'll
generally come up with the cash if your project looks good
enough.

		Jim Kempf		hplabs!kempf