Pensive Street

After the eVoting debacle, the government was found pensively wandering the corridors of Kildare Street, wondering how on earth it could shake off this awful depression and get things moving again.

Actually, I’m joking. If only the government were that reflective, perhaps we’d be in a different position. But they have neither the time nor the right background to make these decisions for themselves, by design. That’s not how the system works: they’re supposed to come to each important issue guided by their own moral compass and the party whip, and when they need domain expertise they turn to the civil service, or commission some report from outsiders which will end up stating what they wanted in the first place. That’s if we even get as far as a report that other people see; with respect to some issues, you never even get to see the underlying justification at all – the two years data retention practice, for example.I remember one day Nick Hilliard rang O2 the mobile provider and got them to send him a list of every cell his phone number had ever been recorded as being in. It was several inches thick, but I don’t remember how far back it went.

But it doesn’t have to be this way. There are a number of examples of how it’s possible to have government and the Internet act well together. Probably the most obvious is the wellspring of the whole thing: academia, from which so much flowed.

Public versus Private

Amongst the many harmful things the financial crisis of the late 2000s brought us in Ireland was the kind of energetic vitriol usually reserved for American political discourse. Here, of course, the vitriol was spilled on the letters pages of The Irish Times and in the pundit shows on NewsTalk, but the enemy in both cases is the same: the public sector. Vilified though it be today, the public sector – and in particular, the Irish universities – were absolutely vital in the development of the early Irish Internet. Without their public-spirited sponsorship, I would not be writing this today. (The case for their primacy is not undermined by stating – correctly – that this sponsorship was sometimes accidental, and sometimes deliberate.)

There’s nowhere better to begin than with the protean and patrician figure of Dr. Dennis Jennings, sometimes called the uncle of the Internet, in that he was responsible for the important decision to use TCP/IP, the now indisputably correct choice of Internet protocol. Dennis is well well-known in the Internet community in Ireland, to the extent of having his own Wikipedia page, but some of the early details, and his views on Internet (and other) governance are well worth your attention.

Dennis did a PhD in astrophysics in the 70s, but was soon looking for a job, and found one with a company called System Dynamics.

Ah. System Dynamics, known as Ireland’s first software house. They weren’t really a software house. They were a body shop. They were a consultancy body shop. But it was at the time – I joined them in 1972 – in 1972 AIB decided to computerize their banking systems and made a very smart decision to go IBM rather than ICL. Bank of Ireland had gone ICL and were much more advanced but had great difficulties. The head of AIB hired System Dynamics to help with recruitment. A hundred new jobs, suddenly.A huge thing for Ireland in those days. There were five hundred applicants and I was given the grunt’s job to sort through them. I remember very well doing that and arriving at – well, this has nothing to do with networking, but did give me tools that stayed with me ever since. That first tool was that if I couldn’t read it, it went into the reject pile. It was clear that the guy had spent less effort filling it out than I was going to spend deciphering it.So much later on, in the interviews for the Head of Systems, they were interviewing the top candidate and Tom McGovern, now deceased, was the Joint MD of System Dynamics was on the interview panel. And he said to the top candidate, “Look, you don’t seem to be very enthusiastic about this job?” And he said, “Well, my boss has also applied for the job.” There was a frantic scrabble after he had gone: “Who’s his boss?”” Eventually in the reject pile we find this application so badly filled out that I had thrown it away. And he got the job.

Dennis learned the trade of programming in System Dynamics, doing work in FORTRAN and a little COBOL. Then mini-computers came in, PDP11s, Novas, and so on, so the game changed quite a bit. This was in the era when a “small” payroll accounting application could cost, say, £30,000+ to develop and install, and computer support was really about working with the largest blue-chip clients to bring the benefits of computerisation directly to them. A far cry from what we have today. Eventually Dennis left and went and became Director of Computing at UCD in 1977, having in the back of his mind a notion of networking which he’d acquired during his work experience.

My first involvement in networking was when I was still with System Dynamics and I think it was ’76 that Michael Purser had gone to Trinity and he had organised a networking conference where all the networking gurus from around the world, particularly the French and the Brits and the Americans. Louis Buzan I particularly remember. I can’t remember if Vint Cerf was there, but certainly the UK guys were there and they were all talking about packet switching and circuit switching and datagrams. And the French were the leading proponents of connectionless datagrams as the way to do networking. I don’t remember who, but I was absolutely fascinat[ed]. I was out in sticks about this stuff and shortly afterwards I went to UCD.

I wondered if Ireland had any non-local connections in those early times, i.e. connections that went between more than just a couple of mainframes beside each other:

UCD had a 9600 baud RJ batch station which they had installed with a 360-50 [mainframe]. That was the highest speed that was available. The banks had multi-plot lines for collection of data to the bank machines. I think those were the only lines. I don’t know of any lines, but certainly UCD with the 9600 bps line was the height of technology and that was the only connection when I joined in ’77. And I got involved in reviewing things. My first purchase in UCD was to purchase 250 kilobytes of memory.Your phone today probably has at least 32 thousand times that size of memory. IBM had some slow mass storage and they could have given me half a megabyte for £25,000 and I bought Memorex, fast memory, 256 kb, for £10,000. And then I bought a terminal, a golf-ball terminal with APL capability, because they were running time-sharing on the 360-50 using APL. And that cost me £2,750. That was a lot of money. But I was [mostly] interested in networking. I think it was in ’79, Michael Purser was a lecturer in it. He was in Computer Science and he came up with the idea that we would do a project together, between Computing Services or Computer Lab or Computer Centre as it was then and Trinity Computer Science to link the two mainframes together. And we hired Jim Mountjoy as the MBST founder. That was ’79, I think. Ahmed Patel was the student that came in to do that. We built this X25 switch, with reverse-triple-X to connect the two mainframe terminal ports and actually have sessions between two Dec-20s. It was called the Irish Universities Research Network, IUN. As far as I know, it was the first piece of networking that was done in the country.

As original and exciting as it was, it was clearly on a small scale, and to be any way serious at all, needed to be expanded. As was often the case with interconnectivity in the early days, there was certainly a general feeling that it was important without necessarily being able to fill in highly specific benefits that were going to be obtained on any funding proposal. Sometimes Dennis had to be a little bit less specific than he might like, sometimes the universities shot themselves in the foot:

Then in 1982 I made a proposal to the Higher Education Authority to link all the universities together with leased-lines to UCD, with a mainframe hub. I was not quite sure what I would do with it but [the key thing] was the link – to link them together. That was shot down by all the other university computer centres because there was no way they wanted UCD at the hub of everything. At that stage, I think Michael Nowlan, who then ran the computers for the Maths department at Trinity, was using UUCP to ship mail [or possibly] Usenet.

A little like the DINX-to-INEX transition, however, the power behind the proposal was too important for one particular implementation to scupper it for too long:

So in 1983, a year later, we made another proposal. I think I led it and with the other universities, to build a network, the Higher Education Authority Network as it came to be known. And funding was provided for that. Work started and the decision was taken to run the UK Coloured Books, because the only set of protocols that actually existed and worked, that we were aware of, were the UK Coloured Books, which were in an advanced state of development in the UK through a programme called Joint Academic Networking, shortened to JANET. The same time, in ’83, I got wind that IBM were sponsoring the extension of BITNET into Europe and I pursued it through my US contacts, a man called Ira Fuchs, who was in City University of New York. Ira got me an invitation so Ireland became the seventh country to attend the first meeting of the European Academic Research Network, which I think was in CERN in Geneva. As a result of which, there was competition to be chair or president and I was favoured over the German competitor.

This was neither the first nor the last occasion that Dennis would find himself at the helm of an organisation. Dennis comes back to the fact that we, garrulous Irish, have a structural advantage in matters to do with European co-ordination:

We speak English but we’re not English! We’re very acceptable. We speak a fairly good English, but also a sort of elliptical English, which is more accommodating to multiple interpretations. We’re not so rigid about it. So yes, we have a significant advantage and very often end up secretary or chair or president of these sorts of things. So, when the actual first link to the world became operational is very vague in my mind. I think that the first links from HEAnet were X25 links. I think there was a long discussion between leased lines and dial-up because in fact, EUROnet had only become operational. This was the experimental, international X25 network, linked across Europe. It only became operational in ’83. [That] was a very active year. I think in ’84 there was some connectivity. And in ’85, while I was away in the States, the first EARN-link, which actually went to Germany initially, not to the UK – it cost IBM an arm and a leg – I think it was probably an X25 link to JANET but I just can’t remember the details. I know there was intense competition between HEAnet, which had a public X25 switching-type mentality, the Coloured Books and EARN, a leased line mainframe-to-mainframe service, which provided by far the best service for Cork and for UCD. Indeed, as soon as that went live, everybody in UCD and UCC had email.

At this stage, though, the motivation for interconnection still seems to be: “You have some computer resources. I have some computer resources. Let’s exchange access to them.”

No. That was the underlying motivation for the Arpanet originally and it never worked out like that. In fact the Arpanet became the platform for running remote access, yes, but also email and mainly for a select bunch of researchers. But also a platform for doing interesting packet-switch things in ALOHA networking - all the DARPA networking, the radio and military portable radio stuff. It wasn’t until 1982 that the Arpanet switched to TCP/IP. I think it was January ’83 when it actually happened, I have the dates written down somewhere. So although the TCP/IP protocols were developed in ’69, in fact it didn’t gain any real traction, in a networking sense, until it was adopted by CSNet in the ’82 proposal. It started in ’83 and they only really took hold with the decisions that I made in ’85 for the National Science Foundation. It was a no-brainer. If you wanted to build an internet, you needed an internet protocol, right?

It certainly did make sense, but at that stage, Europe had one of the largest cases of NIH-ismNot Invented Here syndrome. that has ever been seen. Not to unduly re-open old wounds or anything, but this stage of history was characterised by the Europeans, having their own networking standard called OSIOpen Systems Interconnect, or somesuch. , and the Americans having their own, called TCP/IP. There were immense proxy wars of money and technical talent for much of the 80s and indeed some of the 90s as the European Union in particular, but many research organisations generally, poured money and time into pushing a standard for communication which ultimately failed. Dennis’ most consequential decision, and one of the reasons he will be remembered in history, is the choice that he made to mandate the use of TCP/IP – the simpler, more flexible, but significantly less well specified protocol – across the NSFnet. This essentially ensured that it would ultimately win, and the other things would die out.

But at this stage in history, we are still in the midst of a febrile soup of swirling PhD theses, all suggesting different ways of communicating. Fasctinating diversity, but again, in Europe, the vision was that “there would be a migration across to OSI in due course”, which Dennis realised was an impossibility.

From a very early stage, communication between peopleWhich was at that stage, communication between elites. was an extremely important motivator for growing the network. As alluded to above, though, if you were writing a funding document, as Dennis did for HEAnet, you couldn’t necessarily write that down. Instead, Dennis wrote:

That it would enable access to computers, some level of sharing, an exchange of files and information electronically. The basic tele-type access to resources. Shifting of data around and electronic mail. Now the original motivation for the NSFNet, which was initially called ScienceNet in the design documents until we were sued by ScienceNet CorporationApparently a dial-up mom-and-pop show in Boston. , was to provide access to supercomputers. So, jobAs in, a “batch job” submitted to a computer. supervision. Terminal access and job supervision, controlling of your remote jobs. The insight that I had, well I persuaded people, was that you could build a general purpose network for all science and engineering which would equally well serve the supercomputer users and more importantly that building a specific supercomputer access network would not scale. You couldn’t run links to supercomputer users because in principle every researcher was a supercomputer user. So by definition, you were building a network or networks for all research which would also have all the other benefits of electronic mail, file transfer, access to resources and so on. And in particular data. I’d say the 1982 proposal that I wrote was a single page: “Hey, it seems like a good idea to link things up. You should get some money from the government to do that.” I think the ’83 one was probably a little more rounded. For example, it said things like, “We’ll use it for access to computers and file transfer and data sharing and so on and electronic mail and we’ll use the Coloured Book protocols to put flesh on it.” But it really wasn’t until I’d say ’86, that these things began to come together. I simply do not remember when we had, through whatever circuitous routes, generally accessible email. In ’86 the best network was the EARN network. HEAnet had very low capacity. I think it was 2400 baud, leased line in some cases and 1200 baud dial up in some cases. Again I don’t remember. And then there was, I think there were new 1200 baud modems which were used peerlessly to operate at 1200 baud and I don’t know when any leased line went into the UK for IEUNET. There may well be a fierce argument about who sent which email to whom first and why! All the things I was involved in were academic based. I was not involved at that time in the commercial spin off [world].

Academia was at the leading edge for much of the 80s, and provided (amongst many other offshoots), the largest Internet network in Ireland for at least a decade, the registry for domain names, the leading lights behind the formation of Ireland’s Internet exchange point, and helped to launch the careers of many, including your current writer. That did turn, of course; today we live in an Internet environment which is more dominated by purely commercial concerns than pertained when it was just growing up. For my money, that turn definitely occurred in the 90s, and Dennis concurs, although he sees it as being earlier than I do.

I suppose it became fairly obvious [to me] in 1991/92, when there was a need for domain names. It was primarily the academics of course [initally]. But beyond that, other people also needed domain names. I’d say the key was actually the sale of IEUNET by Michael Nowlan. That was around the same time. But actually the first half of the 90s was tremendously busy. It was 1990 that the RARE/EARN conference was held in Killarney, which was the major … no. Yes. It was 1990 I think. 1989 or ’90. It was the first attempt. It was run by Michael Walsh and Niall O’ Reilly, under my direction. It was the first attempt to bring all the networking people together and chart some coherent way forward. It was quite successful. By 1995 when WBTWeb-based Training, a startup “online learning” company founded from UCD. started, the idea of having a host computer on the campus and being accessible by other people and offering services globally was well established. So yesm those few years, ’90-’95 were tremendously busy. HEAnet moved to leased lines, 64 kilobyte lines in that period. And service became quite good. But I remember going to meetings with Eircom and always speaking from the back of the room and talking about the needs of the academic community and framing those in terms of what the industry would need and [complaining about] the fact that you couldn’t buy fractional E1 and the price of an E1, 2.2 megabit, was 16 times 64 kilobyte and all that, which was absolute madness. Saying to Eircom that we needed the ability to buy fractions of the higher connection speeds rather than the whole of them. No one had a clue what I was talking about. It was very much speaking as the academic, trying to show that this was where commercial networks were going to go.

Dennis would indeed be quite experienced with telling people where they had to go.

IBEC called him up in that period, too, and invited him to talk to them about the Internet:

I spoke to the IBEC telecommunications group and I told them about the Internet and I told them that the Internet was going to be the only network and that all other networks were going to disappear and all computers were going to be connected to this network and all applications were going to be applications on the network using the internet protocols and this world wide web thing. And people looked at me as if I had at least three heads and had no idea what I was talking about.

But he was indeed proved right, although for someone who would traditionally have supported minimal state regulation, he does hold an interesting view on state intervention in this space:

I think that ComReg and all the telecommunications regulators have demonstrated is that it is very hard to regulate an industry that is moving very rapidly. Very, very hard. But at least [ComReg] cut away from the price regulation. I think there are much more important areas for regulation and separation of powers, than in the telecommunications space. Having said that, it is clear now to everybody and certainly it was clear to me at the time that selling off the wired infrastructure with Eircom was a fundamental mistake. It’s also clear to me that intervention is required in key industries by the state to create a platform for the next level of infrastructure. Whatever that is. But I would always argue, and I would argue this for HEAnet for example, is that part of their mission is to do as little as possible. Always do stuff that the industry can’t do, rather than trying to do everything. But I don’t know what is going on in HEAnet now and I am maybe wrong in those comments. So making the transition from intervention because it is needed to letting the market take it over, take over some or all of it over time is across counter-career stuff. It’s hard for people who have established themselves to do that. But that’s what the state needs to be doing all the time.

So, we have shown one potential way for government - albeit things which are only very broadly government, specifically public institutions not under direct political control - to play a positive role in the Internet. There are other ways, of course.

NRENs, and those who fly in them

As an example, we’ve referred above and elsewhere to HEAnet, a particularly positive example of effective interaction between government, academia, and the Internet.As a bit of background, HEAnet is an NREN – a National Research Network. Mike Norris, ex-CTO of HEAnet, commented to me that he would not be surprised if NRENs became much less relevant over the next decade or so, while acknowledging that HEAnet in particular still had a role to play in e.g. the special Internet access network for second-level schools that needs to be run. They have technical know-how and political accountability. But for, say, grid computing, people are just going to use Amazon or Azure or something, and he thought the grid model was not going anywhere. At this stage it makes sense to turn to Dr. Mark Keane, who as ex-chair of SFI and chairman of many committees, is in a perfect position to comment on models that might work or not work. In particular, with HEAnet…

It’s their competence which makes them invisible. All their stuff runs, and it seems to run fine. They seem to have always had a sort of punching above their weight representation in Europe on GEANT and all these other sort of networks. I think they are internationally recognised as doing a decent job with what they do. And they seem to have gotten the forward prediction right, a lot of the time too. I always remember that one of the issues was always that they seemed to be running a lot of headroom on a awful lot of the network at different times but of course, then in the next year, that would be gobbled up. But of course all that environment is just changing so radically.

Even if you think that the role of HEAnet as an infrastructure supplier is not one it is destined to have forever, the expertise that the organisation has developed is still an amazing thing for government to have access to:

The strategy that I have been pushing around is that it is about services and brokering and stuff like that, which again they have developed a lot of expertise in, which is right and appropriate. The issue really here is “Do you want each of the individual institutions to be doing this themselves or do you want a collective body that nationally does it?” And I think the latter. It’s: “Does centralization make sense?” And I think for a place the size of Ireland with the number of universities it has, I think it does. Because trying to duplicate the expertise and staffing across the institutions is just a waste of time.

But why do the governmental institutions seem to have been particularly clueless when it comes to dealing with Internet matters?

So, my experience would have been with two departments. One would have been Enterprise, Trade and Employment, as it was then, who were essentially paymasters for SFI. But in the IEDR, which I was involved in the spin-off out of that, here we were dealing mainly with Marine and Communications. In general, it all comes down to expertise, right? And not only that, but the expertise of one or two people, and I was just complaining about this with regard to Irish Water with my wife last night. We have this generalist civil service model which was probably fine 200 years ago, and is fine most of the time, but the problem is that when you come to things that require deep expertise, then you start buying consultants and you don’t really know if these consultants can do it or not. Or if what they are doing is right. You just accept whatever they tell you. And remember, there’s a mechanism which says, you know, we got the consultant, the experts to tell us and then that takes the responsibility off them for the decision. So, it wouldn’t be true to say that one of the Departments or another Department were poor, they were poor at different times depending on, literally, a particular person being there or not. There was a really good Director General of Marine and Communications who was an engineer originally, I forget his name - perhaps Coughlan or something like that. He was there at the beginning of some of the IEDR events, and then he retired and after that, it just went into a black-hole. You were dealing with these generalists who really didn’t know what was happening. And of course, then they were coming under pressure a lot of the time from industry associations, from particular companies, and you literally had a situation where some CEO would just say a particular thing to them and they’d all be saying it [to us] the next day. That wasn’t a considered response. But they’d be in with you the next day, demanding that special things should be done or something in response. It’s just a sort of expertise deficit. Now, Enterprise, Trade and Employment were very good at a certain time. We and SFI were dealing with Ned Costello who is now – was, I think he still is – the CEO of IUA, the Irish Universities Association, so he left. But he was the main civil servant that we dealt with and he was very technologically competent even though he was an English major.Speaking for those with both Computer Science/Maths & English degress, I must say I resent the “even” :-) But he was marked as well. He was fine to deal with but then of course when things would change, the whole ballgame would change. I once gave a talk to all the senior staff of Enterprise, Trade and Employment on the research environment and the technology aspects. I had this distinct feeling that I was standing up in front of people sometime in the 1950s, you know? You could see they hadn’t a clue. I’m currently doing these R&D audits for the Revenue, and I’ve given them a few presentations as well and its exactly the same.

The above makes it sound like a problem with the people rather than necessarily a problem with the model:

The problem is they need to change the model for how the civil service runs. You actually do need deep experts at certain times. This applies to everything; this is why budgets are over-run, everything is over-run. It’s like saying, you know, “I’m going to put you in charge of a property company now.” “Do you know anything about property?” “No!” That’s crazy. And that’s where the problems have arisen then. People don’t know what you’re talking about and then because, again, they lack the knowledge, they are open to the wrong sort of influence – arbitrary influence, pressure. Policy is … the civil service is always talking about evidence-based policy and the reason they talk about it all the time is that they never do it! They really want to do it, but they never get the time to do it, or politically it just gets over-ruled at a certain point so they don’t really want to do it. But that’s the way it should be done.

The interesting thing is, I think, if you compare what typically happens with the technology related announcements, versus say, health. Health is obviously an incredibly complicated area with a lot of technological developments going on just within itself, but we don’t see the same kind of sustained clangers out of government. Or maybe we do, and we just can’t tell, but that seems unlikely; consider the amount of bad press that health announcements get here if there’s anything even remotely dodgy with them.

Yes, one of the things which happened, especially during the Celtic Tiger years, was there were an awful lot of these expert groups formed. That can be an unbelievable waste of time sometimes in that they produce something and it goes nowhere. The key thing is always, do they have an implementation group – and they often don’t – or a steering committee or something like that [to make it actually happen]. Often the expert group is used to produce a brochure that is handed out but it isn’t implemented. Or even worse, that you get this sort of thing that is often done, where something is set up where it’s all structured in such a way that an answer is given to something that they’ve already decided that they want to do. So, it’s a supporting piece. Now maybe that’s just Realpolitik, I don’t know. But I think it’s bullshit, because it’s not the way to do things properly at all. A lot of that happens.

Mark’s background was as a psychologist originally, and he did his undergrad and PhD at Trinity, but later went on to be Head of Computer Science at UCD, and thence on a number of boards, including the IEDR. As he himself says, “I was in Trinity for eight years: was made a Fellow after three years and then left, which I shouldn’t have done. As everyone would say, ‘How could you possibly leave Trinity?’”:

… the problem I had with it is that I found it stifling, in the sense of this idea that you would enter the place and now you were going to be there till 65. It just hugely depressed me.

So where has government intervention in the technology world worked well, in Ireland specifically?

It can be mixed. For example, what happened with the MIT Media Lab upset me an awful lot. The day it was done, we were criticising it. We partly criticised it because there was absolutely no consultation with the community, no one even asked anyone in the academic sector. The other thing that was in it was it was basically saying, “Well you guys are shite. We haven’t given you any money to do anything…” because SFI didn’t exist at the time, there was no funding. And they were saying, “Well, you haven’t done anything.” “Well, we haven’t been given any money!” Then they were donating 30, 40, 50 million to Media Lab. The history of that, was it was done, literally, over a dinner, between Negroponte and Bertie Ahern. Bertie Ahern was sold on the idea. It was a great big, sort of ad-type thing. It looked good on paper. India shut theirs down six months after it started. I think it took 3 to 5 years for the same thing to happen here. I remember one civil servant saying to me, “Well, it was only 35 million,” you know. It was one of those statements that just sort of rocked me. And I thought, “What could you have gotten for 35 million of investment in the research infrastructure?” You know? Jesus! “An awful lot,” was the answer. But it was a sort of a run-and-jump-at-something. You know the inter-connector, that was another sort of thing that looked sort of white elephantish. They built this huge big pipe going into Citywest. It seemed crazy and it seemed like a bad idea for two or three years because no one was using it and stuff like that. And then, it was really important, you know? There have been things that have been done in the area, like the IDA and people like that; they weren’t really done for especially deep reasons but they sort of got it right. I think the problem is, you have this contradiction whereas on one side government is very ill informed and moving very slowly and doing wrong things and then sometimes just getting it really right, with no difference in the expertise behind the decision. It’s like throwing things at the wall. Some of it sticks! I think a few of those decisions were really good and I guess the problem is how do you maximise them? I think government should take risks on things like that but it’s not done very often. But sometimes when it is done, it really pays off: for example, the inter-connector.

So the key for Mark is really around managing people, and his observation would be that much of the initial conversations that happened when Ireland was trying to grow it’s research base (with the Media Lab, and SFI, and so on) were really conversations around getting prominent researchers from around the world to come to Ireland. That approach can work, but has a number of significant drawbacks:

The whole discourse was about bringing people in. We were going to bring in a Nobel Prize winner and stuff like that. First round, it sort of did that but funded a very small number of people. We applied. A big proposal in. We were all turned down. Most of the people in the country were turned down. They brought in a bunch of people and subsequently it was clear that a lot of them were not very good decisions. I guess if you look back over the nearly ten years, I don’t think the bringing thing [people] in thing really worked. There were a few people that came in and stuck, they were often Irish people who came back. The foreign people that were brought in, it would be interesting to count how many of them are still here. I’d say the number is very small. I often call them carpet-baggers because it was often very large amounts of money involved. Having said that, that doesn’t mean that that was a failure because in a sense, people like that seed a certain level of activity, they affect the other academics around them, that raises the whole bar on how people do things, they create a generation of post-docs who become lecturers, who become professors and that has certainly happened. So, I think the quality of what people were doing kicked up hugely over that period as a result of that and there are one or two, I think, lasting effects. But they are small in number. The DERI thing with Stefan Decker in Galway would be one of the things that stuck and stuck very successfully. Very large German contingent there now, who I think are embedded.

Having looked at two positive models of how government can interact with the Internet, it is time to look at that other great branch of government - the legal profession.