University License fees are short sighted of Wolfram Research
Dave
days and 22 hours ago, but it never appeared. It could be because of the
nature of the post, or it could be because of the fact I used a phony
email address. Anyway, I'll put the post here, with the same phony email
address. Sorry, I don't want any spam.
I mention the use of Matlab here, which on its own can't do symbolic
maths, although there is a toolbox for it. But I was pointing out the
effect the license fees are having on mathematical software. In many
cases, Matlab/Mathematica/Maple and perhaps others too are all just as
effective for many users.
**Sent to comp.soft-sys.math.mathematica on 16th March 2005 at 00:59**
I made the following point to a Wolfram Research employee in an email a
few years ago, but she was not in a position to do anything about it. I
long since forgotten it, but a few years later I thought I'd make it
here. Perhaps even Steven might think about it, should he ever read it!
At the university where I work we have a site license for Mathematica. I
know the approximate cost of this, and feel Wolfram Research are being
very short-sighted in charging so much for the university license.
The department administering the license has to recover the cost, so
they charge the individual users of the product. This is working out at
200 UK pounds per user per year - not an insignificant amount. At the
end of a year, the software stops working.
Although Wolfram Research would have no objections to everyone using
Mathematica inside our university (since we have a university site
license), in practice the cost of the license has to be recovered, so
the department administering it must charge for it. So it is not a "free
for all". Anyone at Wolfram Research who believes that a site license
for a university of 10,000 students means 10,000 students have easy
access to it is very mistaken. It might in some cases be on computers
all 10,000 have access to, but unless it is on their desktop PC, or in
computer rooms they use regularly, it will not get used.
I estimate, based on the cost of the annual license and the cost for a
copy, that only 1% of the university are paying for the right to use it,
despite us having a site license.
Now compare that to Matlab, which for many will do the same job.
We don't have a site license for Matlab, but a license for a very large
number of copies. It is "effectively" a site license, as the number of
users will always be far less than the number of licenses held.
The cost per user for Matlab is 100 pounds per copy, but this gives us
free upgrades for a period of 4 years, making the annual cost an average
of just 25 pounds per year. At the end of the 4 years, Matlab will not
stop working, but continues to function, but without further upgrades.
So in summary:
Mathematica = 200/year. Software stops working after 1 year.
Matlab = 25 pounds/year with free updates for 4 years. Software
continues working after the 4 years.
So Mathematica is costing 8x as much over a 4 year period, and even more
if you consider Matlab is not time-limited.
I'm *not* saying Wolfram are changing 8x what Mathworks are, but the
higher cost for the Mathematica site license pushes up the cost for
those that pay to use the software, which means less people use it,
which pushes up the price further. It is a vicious circle.
Is it any wonder that my own department, that used to use Mathematica
heavily, has now switched over to using Matlab? I'm one of the few (if
not the only) person to use Mathematica. My boss wants me to stop using
it, as anything I write will not be usable by anyone else in the group.
I suspect I'll have to switch myself - despite the fact I paid for "a
license" yesterday.
The mathematics department still use Mathematica, but I learned
yesterday they seem quite happy to use Maple instead. So I suspect that
when the Mathematica site license expires later this year, there is a
good chance the site license will not be renewed, but one obtained for
Maple instead.
I can't help feel this is so short sighted of Wolfram Research. I would
think they would be keen to almost give Mathematica away to
universities, knowing the short-term loss on these license fees would
give a long-term gain when students go out to work and buy software.
I've no wish to start a Maple/Mathematica/Matlab war. They are all
different, all have their strengths, but for many users, it is fair to
say any of the three will do what they need.
I will not be involved in the negotiation of any new license - just
making this observation as a reasonably senior post-doc scientist.
PS
Just as a slightly off topic comparison, I use a Texas Instruments
Digital Signal Processor. Texas give free seminars on using their DSP
chips. For student projects they will usually give away the DSP
evaluation kits (this is hardware, so has a real cost to make further
copies of). I once went to a seminar and the guy said basically "the
answer is yes if you want something free for academic use. Now what do
you want?" He said they would draw the line at very expensive hardware
for undergrad projects, but within reason you can get hardware free in
Europe.
I know one academic who got 20 TI DSP evaluation kits for free. He had
to write a short justification for this, but got them.
TI are well aware that if they introduce students to their range of DSP
chips, those students are going to want to use them rather than DSP's
from Motorola or whoever else makes DSPs.
Richard Fateman
better than Stephen Wolfram and his presumably well-paid
market-droids.
I am sure they have considered many different models and
have decided that this is the way that makes them the
most money. Maybe not the most friends.
My guess is that the most lucrative sales are to universities,
and WRI is not about to give it away.
Dave
> You are supposing that you know how to market Mathematica
> better than Stephen Wolfram and his presumably well-paid
> market-droids.
Although if you read his book 'A new kind of science' you soon realise
he is very self-oppionated, with just about the only science book I have
read where the author keeps referring to 'I' all the time. Hence it
would never surprise me if Stephen Wolfram felt that high prices to
universities was the way to go, then that would be the remit of the
marketing people.
> I am sure they have considered many different models and
> have decided that this is the way that makes them the
> most money. Maybe not the most friends.
I'm sure in the short to medium term it does make more money, as it
would take a few years to get universities to heavily adopt Mathematica
and for the graduates to be in a position to specify software when they
are in industry.
> My guess is that the most lucrative sales are to universities,
> and WRI is not about to give it away.
In the science + engineering fields I often see jobs advertised wanting
Matlab skills, but never Mathematica. (I guess jobs at Wolfram would be
an exception!) Looking on the newsgroups, I currently see
maple - 102 messages
mathematica - 388 messages (I do realise this is moderated, which does
not help matters either).
matlab - 2368 messages
I can't be bothered to try to sort out the academic posts from the
non-academic ones. But both these facts (jobs and newsgroups posting)
would indicate Matlab have a much larger share of the mathematical
software market than Mathematica. I wonder if there is any correlation
between this and the fact Matlab is provided to universities (well at
least where I work) at a far lower price.
A quick look on the websites of Wolfram Research and Mathworks shows
Mathematica was launched in 1988, and Matlab in 1984. Wolfram Research
claims there are 7 million users of Mathematica
http://www.wolfram.com/company/background.html
but I wonder if they count users as universities that have site
licenses. Sure, we have a license that would allow 10,000 users to use
it, but in practice since the license fee has to be recovered, in
practice the number of users is far less.
I know some here take the opportunity to slag Wolfram Research off all
the time. I'm not doing that - I like Mathematica, and despite having
access to Matlab too, I prefer the former. But I realise that soon I
will probably have to switch to Matlab, since
a) The university is unlikely to continue the site license with Wolfram.
b) From a point of view of a CV, Matlab skills are more desirable.
Nasser Abbasi
>
> So in summary:
>
> Mathematica = 200/year. Software stops working after 1 year.
> Matlab = 25 pounds/year with free updates for 4 years. Software
> continues working after the 4 years.
>
> So Mathematica is costing 8x as much over a 4 year period, and even more
> if you consider Matlab is not time-limited.
hi;
To be a little fair, Mathematica for students version costs $139
in my school, and the Matlab for students version costs very
close to this amount (I do not remember the exact amount,
may be $129).
Why is it important to have a site license when the cost of
individual student copies is low?
Nasser
Vladimir Bondarenko
D> Wolfram Research claims there are 7 million users of
D> Mathematica
D> http://www.wolfram.com/company/background.html
A tiny correction. I believe, somehow 'several' turned
into 'seven' :)
http://www.wolfram.com/company/background.html
WRI> Mathematica Version 1.0 was released on June 23, 1988,
WRI> and was immediately lauded by the scientific and technical
WRI> community, as well as the media, as a dramatic advance.
WRI> Within months, there were tens of thousands of users around
WRI> the world, and today Mathematica's reach has grown to
WRI> several million enthusiastic users around the world.
^^^^^^^^^^^^^^^
Best wishes,
Vladimir Bondarenko
http://www.cybertester.com/
http://maple.bug-list.org/
http://www.CAS-testing.org/
Dave
> Dave <nos...@nowhere.com> wrote on Sat, Mar 19 2005 8:33 am
>
> D> Wolfram Research claims there are 7 million users of
> D> Mathematica
>
> D> http://www.wolfram.com/company/background.html
>
>
> A tiny correction. I believe, somehow 'several' turned
> into 'seven' :)
>
> http://www.wolfram.com/company/background.html
>
My mistake.
Dave
Student copies can not be installed on a campus PC,but only on the
student's personal computer
http://store.wolfram.com/view/app/mathforstudents/
Unless the post-doc researchers and lecturing staff use the software,
the undergrads will not be encouraged to use it.
I just looked at the student numbers where I am and found they were
about twice what I thought - there are around 20,000 students. *IF* I
have the correct figure for the license, that is about £1/student
(~$1.80/student) - not a lot you might say, and even less when you
consider staff can use it too.
What would happen if it was free?
The usage would dramatically rise. Wolfram Research would only need to
sell 100 extra copies/year at commercial price to offset the loss of
license fee to the university. If an average student stays 5 years, that
means there would need to be a 0.5% increase in graduates leaving the
uni and buying copies at work. That will not be achieved in the first
year, but after a number of years, the popularity of the package would
rise and so the number of copies bought at commercial prices would rise.
Currently based on the price we pay for a copy if we want to use it
(£200/year), I think less than 0.5% of students+staff are paying for
copies so they can run it.
You might reasonably argue that if we have a site license that allows
20,000 students + staff to use it, then we should allow anyone to use
it. I guess I would tend to agree with that, but the fact is that is not
happening, as the license fee has to be recovered. That is not my
decision, but a fact of life where I work.
I don't have any financial input on this, but I might try to change
this, as it does seems a silly situation.
Richard Fateman
><snip>
>
> What would happen if it was free?
>
> The usage would dramatically rise.
My guess is that you are wrong here.
This assumes a kind of elastic market that probably does
not exist. By your reasoning every piece of free software
would be used by every student. If Mathematica dropped its
price, it might sell about the same number. The number
is bounded by the "math geek population" more than the
"population with $20 to spend".
In reality, most people in their jobs do not even need
Matlab (whose numerical-only capabilities are much less ambitious
than Mathematica or Maple or ...). The employer is
much more likely to provide a license for Microsoft
Powerpoint or Excel to each technical employee.
It is a relatively small market, and "several million"
is probably inflated. Perhaps the sum of enrollments
in calculus classes/labs in all colleges with
site licenses over the last decade. Certainly not
a count of people who used Mathematica seriously.
RJF
Dave
> Dave wrote:
>
>
>> <snip>
>
>
>>
>> What would happen if it was free?
>>
>> The usage would dramatically rise.
>
>
> My guess is that you are wrong here.
>
>
> This assumes a kind of elastic market that probably does
> not exist. By your reasoning every piece of free software
> would be used by every student.
No, I am *not* claiming every bit of free software would be used by
every student. How you infer that from what I wrote I do not know.
Whilst I might guess some historians studying 17th century literature
might use Mathematica, the product will only really be used in any great
number in science, engineering and maths courses, with lesser (if any)
usage on other courses.
But providing it free would likely mean it would likely to be taught in
just about every physics, engineering and maths course. I know it is
used in computer science too.
I know that several years ago a decision was made in my department not
to contribute any more toward the cost of a university site license for
Mathematica, on the grounds of cost and the number of users. That
effectively reduced the number of users to about 3 in a department of
150 or so staff. There must be 50+ staff using Matlab.
The uni still has a site license - but since we don't contribute to that
cost, we don't use it unless we buy a copy, as I have done.
To a large extent, the number of undergrad students using a product is
set by the number of lecturers/post-docs that use the product. And that
is set to a large degree by the cost of the product. If two products are
each capable of doing the required task, the cheaper or one best known
to that individual will generally be bought.
> If Mathematica dropped its
> price, it might sell about the same number. The number
> is bounded by the "math geek population" more than the
> "population with $20 to spend".
But where I work at least, it is not someone with $20 to spend, but (for
staff like myself) someone with around $360/year to spend. I know with
100% certainty that we no longer use it much due to the cost. Nobody is
too keen either on software that stops working after a certain date. No
longer being able to get support and updates is not such an issue, but
it is not too tempting to buy something that you know will stop working.
I can't say how many more users there would be, but believe it would be
significant.
And I know there is talk of not renewing the site license any more. With
around 20,000 students (many on mathematically based courses), that
would be a loss of Wolfram Research's potential future sales.
> In reality, most people in their jobs do not even need
> Matlab (whose numerical-only capabilities are much less ambitious
> than Mathematica or Maple or ...). The employer is
> much more likely to provide a license for Microsoft
> Powerpoint or Excel to each technical employee.
Matlab has a toolbox for symbolic maths, which still does add much to
the cost. Someone told me the symbolic toolbox for Matlab is a cut-down
version of Maple, but I don't know if that is true.
> It is a relatively small market, and "several million"
> is probably inflated.
I suspect that too.
car...@colorado.edu
How many toolboxes does your campus Matlab license cover?
As for users, here are rough guesses for my campus - I get
these from serving in college-level IT committees that
are informed about licenses, etc:
Matlab: 1500-2500 students/yr, mostly engineering instruction
Mathematica: 300-500 students/yr, mostly applied math &
physics instruction
Maple: unknown - not used in courses except as matlab toolbox
Serious users (research) 200 Matlab, 50 Mathematica
Total campus students: 29000. Engineering: 3500. Sciences:
(math, physics, chemistry ...): 6000
There are 100-150 similar campuses across the US so for the
US-total append 2 zeros. Doesnt reach to 1M users though.
I agree with Professor Fateman that the market is small, but
it is slowly growing.
Dave
>>>It is a relatively small market, and "several million"
>>>is probably inflated.
>>
>>I suspect that too.
>
>
> How many toolboxes does your campus Matlab license cover?
I can't give you an exact figure, as it varies from toolbox to toolbox,
but I'd guess 500-1000.
The Matlab licence allows home usage, but the procurement department
administering that said
"Yes it does, but you are still going to have to buy a license if you
want to use it". So I paid £100 (~$180) of my own money to run a copy at
home on my Sun - since I wanted it on a Sun, I could not get the CDs or
license key any other way.
> As for users, here are rough guesses for my campus - I get
> these from serving in college-level IT committees that
> are informed about licenses, etc:
>
> Matlab: 1500-2500 students/yr, mostly engineering instruction
> Mathematica: 300-500 students/yr, mostly applied math &
> physics instruction
> Maple: unknown - not used in courses except as matlab toolbox
> Serious users (research) 200 Matlab, 50 Mathematica
>
> Total campus students: 29000. Engineering: 3500. Sciences:
> (math, physics, chemistry ...): 6000
>
> There are 100-150 similar campuses across the US so for the
> US-total append 2 zeros. Doesnt reach to 1M users though.
> I agree with Professor Fateman that the market is small, but
> it is slowly growing.
>
Whilst this is slightly off-topic, I'm interested in how other
universities manage site licenses for software. Where I am the process
seems to be.
A) The procurement (the most useless department of all) buys the site
license. Thankfully the Mathematica license is *not* done by procurement
any more (they made a real mess of it) but by the computer science
department.
But I gather it will go back to procurement's hands (if renewed), as CS
are making a loss on it, which they don't want - procurement are
expected to make a loss. (Something like that anyway).
B) The procurement (or CS in the case of Mathematica) department then
issues CD's and a "license certificate" (nothing more than a bit of A4
paper produced internally) on the receipt of a "license fee". A license
for a whole department will cost more than a license for one user.
This "license fee" we pay to CS or procurement is set to recover the
cost of the site license paid to Wolfram Research, or whoever produces
the software.
This means only people that use the software pay for it. If you don't
pay, you don't get issued the disks, passcodes or a bit of A4 paper that
says you can use it.
Hence we have two equations.
cost_per_user=license_fee/number_of_users (1)
number_of_users = f(cost, how general purpose it is, etc) (2)
I guess the English Department would not be keen to contribute towards
the Mathematica site license if they never use it, no more than my
department would want to contribute to the cost of some software for
inorganic chemistry, since it is not something we would use.
However, this leaves a lot of "gray areas" where someone might well use
some software (like Mathematica) if it is readily available to them, but
since it is not, they don't use it. Many in my department did use
Mathematica seriously, but no longer do because of its cost.
Since we have a Mathematica site license, in theory anyone can use it
and I'm sure Wolfram Research would like that. But in practice it is
available to few, as the license fee we pay internally is too high. This
has to be high to cover the cost paid to Wolfram Research and the fact
there are few users.
It's difficult to see a solution to this one.
PS
Just to illustrate how useless our procurement department were with
Mathematica, I found out by accident one day from a Wolfram Employee
that our site license allowed use at home. Nobody had ever told us this.
So I asked for a SPARC license. Originally Wolfram Research said a Sun
was not a home computer, but did this as a one-off, which I appreciated.
In fact, our current site license is much more restrictive on home
usage, as the computer needs to be owned and managed by the university.
car...@colorado.edu
Our licensing process is rapidly changing. It has become more
decentralized and flexible. Not by design: it responds to market
forces. Probably the same change is going on at other schools. There
is no central procurement. Any unit (campus, college, department,
institute, center) with their own IT budget can negotiate with vendors,
and try for best deals.
Ten year ago campus wide licenses were the norm, and were administered
by Information Processing (now called Information Technology Services).
They still do negotiate on software of campus wide utility - for
example Office. The trend is to obtain academic discounts from bidding
suppliers rather than direct-from-vendor licenses. For example I
purchased Office 2004 for my Macs for $55 with a designated supplier.
At this level there are no constraints on "university machines" versus
"home machines". With everybody using laptops, the distinction makes
no sense.
Licenses for programs with wide instructional use such as matlab or IDL
are purchased by departments or colleges. For example the College of
Engineering negotiated a fairly generous Labview license - anybody with
the proper ID can get a CDROM. It is not valid for people outside the
college.
Departments and centers tend to buy multiple user, year-to-year
renewable, server licenses for use in labs, but those are restricted
to clients. CAD and simulation software such as FEM codes tend to fit
that framework.
In summary the trend here is: you buy only what you need, for the
people that will use it, using whatever leverage you can muster. The
days of blanket university-wide licenses are fading. They cost too
much to administer and keep track of. What is the use of matlab for
(say) the Religious Studies department, a music composition program
for Sociology, or Maya for Geological Sciences?
Jon McLoone
> and I'm sure Wolfram Research would like that. But in practice it is
> available to few, as the license fee we pay internally is too high.
This
> has to be high to cover the cost paid to Wolfram Research and the
fact
> there are few users.
>
> It's difficult to see a solution to this one.
There are a number of ways to license Mathematica in the UK, some of
which can work out to much lower unit cost than you suggest.
Without knowing which university you are with, I cannot comment on the
choices which they have made or the way in which they choose to cost
recover. If you would like to contact me directly, we can try and work
with your site administrator to see if there is a more optimal
solution.
> PS
> Just to illustrate how useless our procurement department were with
> Mathematica, I found out by accident one day from a Wolfram Employee
> that our site license allowed use at home. Nobody had ever told us
this.
> So I asked for a SPARC license. Originally Wolfram Research said a
Sun
> was not a home computer, but did this as a one-off, which I
appreciated.
>
> In fact, our current site license is much more restrictive on home
> usage, as the computer needs to be owned and managed by the
university.
Incidentally, this looks like a mis-communication too. Our home use
licenses do not require that the university owns or manages the
computer.
Dave
> There are a number of ways to license Mathematica in the UK, some of
> which can work out to much lower unit cost than you suggest.
>
> Without knowing which university you are with, I cannot comment on the
> choices which they have made or the way in which they choose to cost
> recover. If you would like to contact me directly, we can try and work
> with your site administrator to see if there is a more optimal
> solution.
I'll reply to you off the newsgroup.