Re: Initial survey... some thoughts

[Roger Dingledine <arma@mit.edu>]

In message <Pine.SOL.3.96.980618164010.20816C-100000@kanto.cc.jyu.fi>, hvirtane@cc.jyu.fi writes:
>On Wed, 17 Jun 1998, Roger Dingledine wrote:
>
>> question from "how can we make linux better?"
>
>1)
>>From the developmental point of view I liked to see it from this point: 
>
>We need some new kind of people to *test* linux to get new ideas how and
>which way to develop it. 
>
>One possibility is to find out, which kind of people in general use
>computers and what they do with it... But that is already probably known,
>because we can gather ready made facts, about how many of different
>operating systems and applications are sold worldwide? 

I'm not aware that any statistics like this are around that pertain to
Linux -- it isn't an issue of how many are sold, because the Linux ones
are freely downloadable (and I guess piracy might be an issue for commercial
ones?) Also, whether somebody purchased a word processor doesn't tell me
how much they use it, what they use it for, and what they like about it.

>So, one possibility is to find out these kind of users' profiles and to
>show that linux can do the same and something more and cheaper and better? 
>We coulkd perhaps try to ask, what is missing in people's systems?

Yes. Ideally, we would ask the questions in such a way that we aren't trying
to force people to realize that Linux is cheaper and better. From the
OS-neutral answers that people give, we hope to be able to draw the
conclusion that Linux actually can do what they people want (or can't,
and if it can't why it can't).

>(To take in consideration the kind of applications I personally mainly
>use,
>in my opinion at present it isn't a question of a wordprocessor being
>better or worse (for example word or wordperfect) but 'taste' and PRICE. 
>Earlier I got used to wordperfect, but it was easy to shift to use msword,
>because WP didn't work on this old mac, but msword did. I can easily do
>the same things with both of them. (Earlier I used an IBM-clone at work.)
>(The person, who instructs the users of wps here at this university is
>said to instruct people only in WP's use now, because it is now so much
>cheaper; one year ago it was otherwayaround.)) 

While it's true that most of the commercial word processors are identical
in terms of technical functionality, as you indicate there are some
non-technical differences that are extremely relevant. Our goal is to
figure out which features people actually care about. That is, how much
can I take away from a word processor before you say "Wait a minute, I
can't write my memo in that anymore!" Also, I want to try to figure
out how much the differences matter -- whether it's price, or tech
support, or which hardware it will run on that is most important to most
people.

>2)
>The other way to look at it could perhaps be to take the linux more or
>less as it is now. See, if it can do the things people do in general, make
>different packets for different user profiles, find a group of new people
>to give it a try... I mean we don't need to make it much better, but easy
>to take in use... I have something quite similar in my mind as RH or
>suse, but basic installation should be easier as well as the
>documentation. 

Freelinux and independence are working towards "easier install" goal
currently.  There are several people working on "better installers" as
well (see http://www.seul.org/dev/install/ for a draft of an install
specification that's currently stalled due to lack of workers. Also,
there's plenty of discussion about that in the dev-install archives.)

Documentation is a different matter, though. I'm sure there are people
out there generating good documentation. I'd love to find them and
coordinate them, though. (If only I had time....)

>Linux can already beat the others in one important area: price. Perhaps
>that is already enough given we make it easy enough to install by normal
>end-user. 

I suspect market share will remain a huge issue for a long time. But I
guess maybe that's less of a problem outside the US?

>But: 
>the problem is that even many people whom I know, who have
>used a computer 10 years as their main working tool have never installed
>any os or even application themselves in their computer. It is an other
>person (the person in charge of all this working-place computers) who has
>done all the program installations. Probably the only way to spread linux
>here is to get it on some new computers they buy.  
>
>All those people, whom I know to use a computer at home, bought the
>machine with an os (you know which oss)  already installed. Why should
>they change it? They already paid it and and it is easy to buy more
>applications if they need them... 

This is another good point. That's something that the linux advocacy
effort (another not-very-coordinated-group, ugh) is working on.

>So the 'normal end-user', who is capable and willing of installing an os
>her/himself is not existing? 

I think such a user does exist. Perhaps that user doesn't qualify as a
"normal end-user" currently, but certainly linux is becoming more and
more simple to install and use, so it will get there in time. (At the
same time, the linux advocacy efforts are doing their part to remove
some of the stereotypes about linux, so more people will be willing to
try it.)

>(I just asked a friend (, who has probably a little bit
>more experience with computers than I,) if he wanted to try my RH5 on his 
>machine... he doesn't have a cdrom-drive... forget it... Probably the
>installation media is important as well...)    

Without a cdrom drive or a good net connection (local or global) these
days, it's tough to install a big operating system+apps. I don't see a way
around that. 

[bullet list of ideas snipped]
>I think that all the properties listed above are important. And it is
>perhaps not the right question to ask, which one is the most important or
>how important it is. Probably it is better to try to do things that way,
>that the user can select himself an user profile considering networking
>and applications and such things, which are yes/no choices. I mean such
>things which are already available with linux (like the existing
>distributions already do.) But probably those kind of properties should be
>made possible to install by just answering 'yes'... In my opinion all the   
>things, which some developer invents can be important to someone. It is
>perhaps difficult to know beforehand how important something is, before
>people have used it? 

At first I disagreed with you here, but I've thought about it a while
and I think you might be right -- in order to give the people filling out
the survey a fair shot at having an opinion on a question, they have to
actually have tried a couple of the options to know which they prefer
more. Perhaps an "i don't know" should be a valid option. On the other
hand, I'd like to word all the questions so they don't need to have
experienced the situation to have an opinion on it. That is, it's
theoretical enough that they don't need to actually try it to know what
they'd think.

Also, a lot of the questions are things that the user has already
experienced with whatever operating system he/she currently uses -- things
like PNP recognition, cost, speed, etc.

It would be nice to give them software and see what they think. Actually,
we're already doing that: they have an operating system (either we gave
it to them or somebody else gave it to them (ignoring price issues)), and
we're assuming they've used it for a while. Maybe I think you're wrong
after all. :)

>The other way to ask about things is to ask, where the problems are with
>people's systems. To ask, what people are doing with their machines and
>what more they wanted to do and try to give them solutions to their
>needs?

I think that coming up with solutions has to happen after we've written
the survey, given the survey, and tried to assess what their needs are.
Certainly one of the goals of the survey is to be able to do that, yes.

--Roger