RESPONSE TO THE "DISCUSSION PAPER" OF THE PROVOST'S TASK FORCE ON ACADEMIC COMPUTING Stefan Mochnacki, Dept. of Astronomy. 21 Feb. 2000 My response to the Discussion Paper is based on 33 years of computer usage, 18 of them at the University of Toronto. I use computers in all aspects of my work, and during my time as an astronomy professor I have been involved heavily in computer administration, including the design and construction of three initial departmental networks. In addition, I served several years on the Council and Executive Committee of the U of T Faculty Association, and likewise designed and still help maintain its computer network. My perspective is the rather unusual one of a science professor who also gets involved in the nitty-gritty of hardware and software, and has been involved with wider societal matters as well. Since the Discussion Paper is quite long, I will address its sections sequentially, using the formatting of the PDF version for page references. I use "IT" to mean "Information Technology", and "IP" to mean "Intellectual Property". The response represents my personal opinions and not those of the Department of Astronomy or any other group. However, I have circulated earlier versions of the response to various people for comment. PART II Section 1. Page 7: Existing print material may appear in digital form quicker than expected. For example, essentially the entire American astronomical journal output since its beginnings in 1895 has been scanned and is available on line. "Support for end-users": We especially need TRAINING of faculty and other staff. Courses and certification should be available. Most of us really know nothing about truly effective use of IT in teaching. The recommendations on p.10 are welcome. Section 4, p.14: Recognition and Reward of Use of New Technologies Use of IT could be reflected in the teaching assessment forms filled out by students. Dialog is needed with UTFA and CUPE 3902. There are academic freedom issues here: should not faculty have absolute discretion over use of IT? Is the use of IT in teaching sufficiently well established to warrant its being rewarded per se? This use of IT needs a thorough debate, rather than being mandated as if in a business corporation. Experience at York and elsewhere has shown that unless the integration of IT into teaching involves all parties as willing partners, severe institutional dysfunction (e.g. strikes, dissension) follows. The already announced conferences will help, but the issue needs to go through the collective bargaining process to ensure harmonious progress. Section 5. p.15: Distance education: It is extremely important that we develop tools which are platform independent and can be built on over a long time. Let's not reinvent the wheel! This means using open source throughout (e.g. Apache, PHP, Zope etc.) Section 6. Intellectual Property rights. It is a very wise approach to leave content IP rights with the author. The same should apply for tools, which could encourage people to use the General Public License (GPL), making such tools freely available to a wider community. PHP started that way here at U of T. Perhaps the University should reward people who develop useful things and place them under the GPL. The free adoption of such GPL'ed products by other institutions would bring benefits to U of T. Part III, section 1, p.18 "A Decentralised system with Central Coordination". Systems Support Group: it should include a few faculty who have long-standing experience in system administration. These people provide CONTINUITY, because there is considerable turnover of sysadmin staff in the departments and new sysadmin staff have to be trained by them, and COMMUNICATION between sysadmins and academic policy groups. (e.g. in Astronomy, Charles Dyer provides this function at Scarborough, and I do at St. George.). If a department has just one sysadmin, when the sysadmin leaves a new one has to be trained, usually over a much longer period of time than any overlap between outgoing and incoming sysadmins. The permanent tenured faculty are the only people with the long-term institutional memory needed to train the new sysadmin. Regarding communication between sysadmins and faculty policy makers, I note that the previous five-yearly Academic Computing report (ref. below) took a rather different direction after I alerted its Task Force, to the existence of the "ut-admins" group: a meeting was called within two hours. Faculty also act as a catalyst for sysadmins, an understandably conservative group, to consider new directions. The net result is more practicality and more innovation, a win for everybody. "Incidence..." ... should be "Incident Response Team". (p.18). I STRONGLY urge that this service be CENTRALLY FUNDED, without billing of departments. If billed, we'll just continue the same old "do it ourselves" way. Perhaps it should consist of "volunteers", highly experienced sysadmins and technicians drawn from CNS and the departments. Sort of a "Volunteer Fire Brigade", with some compensation to the departments from which they come. One cannot predict the need beforehand, so budgeting would present a severe problem to department chairs unless we were to have some sort of "insurance" scheme. A per-incident charging scheme would not encourage use of this service, but an "insurance" plan could finance it properly. Section 4. Equipment Renewal Over the last 20 years, I have found that the renewal cycle is 4 years for hardware, 5 years at a painful stretch. Newer, essential, software just keeps getting bigger and more CPU-intensive. Older equipment is unsuitable for use in the University, although some components can have a longer useful life (though their resale value is negligible). The old equipment must be removed from the University; in fact, until we made a determined effort at disposal, we were getting swamped by discarded equipment cluttering Astronomy spaces. Obsolete equipment over 5 years old generally cannot be upgraded to satisfy security needs, and thereby becomes a serious security hole, although in some limited cases this is not so (e.g. a 486 box acting as a print server). A good thing is that IT systems (hardware combined with software) have actually gotten cheaper as well as much more capable. An aggressive renewal policy (assume 4 years for everything, say), would ensure that the use of IT is not held back by a particularly low common denominator. Students, in particular, should not have to use painfully obsolete systems. Old equipment is a poor recruiting tool for students. I would recommend that all departments have a master renewal budget plan based on a 4 year cycle, so that money can be sought and allocated well in advance. All computer equipment should be stricken off the University's books after 5 years, so that the threat of paperwork does not discourage its disposal. Section 5, "Open Source": I argue that we should keep strictly to OPEN STANDARDS regarding protocols and formats, so that all essential operations are platform-independent and can inter-operate. OPEN SOURCE should be the default for everything, with free software used as much as possible. PROPRIETARY software should only be used for specific uses; nobody should be forced into adopting a particular computer architecture because that's the only one supporting a particular infrastructure application. We must avoid the blandishments of vendors offering normally expensive proprietary products for free, which would tend to lock people into a particular closed architecture or environment, to the vendor's ultimate benefit. This is a Trojan horse to be refused. Unfortunately, a number of American universities have signed such deals. Some people, especially students, think that the systems they learn to use at the University should be the same as those they will use in the workplace. Actually, the user interfaces of open (or free) systems are now very similar to those of the dominant proprietary ones, and in many cases they are the same. Conversely, the use of open systems is rapidly increasing in the private sector. In the long run, the University can save money and gain capability by migrating to largely open source systems. Sections 6 & 7: AMS, ROSI, user authentication. SAP now support open architectures (e.g. Java, Linux). I suggest that AMS and ROSI likewise fully support all client architectures. Why does the University feel itself bound by US export laws? Perhaps an in-house or made-in-Canada encryption package is needed. Perhaps we need to adopt a better Web browser technology, with our own security add-ons. The PKI idea sounds good: why not contract local programmers to do the job? (Henry Spencer is an obvious candidate to lead such an effort). Again, I suggest investing in the development of open source solutions rather than buying closed implementations from outside vendors. IV: Faculty and Student Access. I would favour cheap "Web appliances" over laptops for student use, combined with home workstations. I discuss this in Section 5 below. Section 2, connectivity of residences. The costs must be made MUCH less than WAVE/Sympatico rates (approx. $40/month). With the existing network on campus, our costs can be much less if handled internally or through a local contractor. A U of T CLIENT WORKSTATION modified Linux distribution should be produced to make installation and security simple (this is being done at US universities, e.g. Michigan). Very clear policies regarding the operation of servers in residences should be enacted BEFORE the connections are built. (The use of DHCP and firewalls could eliminate most problems). It may be wise to allow students some freedom to experiment so as to permit the development of hands-on Internet expertise. Students must be free to use open source systems in dormitories. A good dormitory computer policy would help in student recruitment. Section 3. Web-based e-mail. Absolutely! Long overdue. The University could use such a service to advertise extension courses, special events and so forth. It would be a great way to keep alumni in touch with U ot T (why should they go to Hotmail/Yahoo/et al.?), and would help to retain a community which is crucially important for fund raising. Section 5. Portable computing. We should NOT make laptops compulsory! They are TOO expensive, and not sufficiently capable. Students carry too high a cost burden already. Also, there is a security/theft problem with them, and they would get too much rough handling. Except in certain specialised computer-oriented courses, laptops are disruptive in class, and too dependent on students' typing skills for efficient use. They simply distract from the teacher-class discourse. Instead of taking notes during class as if in dictation, students should use the instructor's Web site to get precise information later (or beforehand); class time can then be used for gaining UNDERSTANDING, a much more efficient use of everybody's time. The big IT display system in the classroom is perfectly adequate for conveying what the instructor wants to convey (and there's still a blackboard, chalk and the old-fashioned O/H projector...) On the other hand, I can see where low-cost "Web appliances", perhaps wireless, may very rapidly come into widespread use (the Crusoe processor comes to mind). A $300 device would be much more accessible to students than a laptop costing ten times as much. Such an appliance would be much better than a laptop for INTERACTIVE class use. This is a very rapidly evolving emergent technology. The wiring of every seat in a classroom is therefore redundant, except in certain cases (e.g. Faculty of Management). I think the Task Force is misguided in believing that portable computers will prevail over desktop PC's. A more likely scenario is that both portable and desktop PC's will be replaced by low-cost portable "pads" and centralised application servers. Only specialised "power" users (e.g. research students, faculty and administrative staff) will require full PC's on campus, whether portable or desktop. (Home-based workstations would be used for data transfer and long-term storage; a laptop is not a secure place for storing valuable data in large quantity.) There are simple examples of this approach in existence already: Web-based e-mail (e.g. Hotmail, Yahoo), and centralised word-processing (e.g. NSERC's application Web site). There has been a lot of speculation about this trend in the trade press, and recent speeches by industry leaders at computer shows are quite clear about it (e.g. McNealy, Torvalds). Prototypes have been demonstrated. 6. WEB pages for individuals. I think it is very important that people can make Web pages and install them regardless of what platform they used to create them. The underlying WEB server software and its plugins should be open source and standard. 7. Portals This is an interesting section. However, we must ensure that departments and faculty remain free to operate their own Web servers, and have them linked as they see fit by more central Web sites. The existing recent Web policy is quite good in this respect, but freedom of expression (within the law and the appropriate use guidelines) must be respected. My Department (astronomy) has always used its Web site as a simple portal, and I find it effective even off-campus. I am sure there are many others at the U of T. REFERENCES: I have written earlier on some of the topics discussed above: http://www.caut.ca/English/Bulletin/98_dec/comment.htm http://utl2.library.utoronto.ca/disk1/www/documents/bulletin/feb8_99/forum.htm ----------------- A few links and other stuff can be found at: http://astro.utoronto.ca/~stefan/ ----------------- The previous (1995) Task Force on Academic Computing report is at: http://www.utoronto.ca/provost/tfacnm/draftFeb12.htm