Click here for a link to an article from the Lasa publication Computanews, on the some of the recent developments around the electronic provision of advice services.

I am interested in what people's thoughts are around e-advice in particular - do you think that a computer programme can realistically carry out the role of an advisor? As budgets and funds are increasingly squeezed, will developments such as Project Eagle be the way forward for advice provision in the 21st century? Could the internet be the method for overcoming access issues in rural areas?

Please post your thoughts here, and depending on the interest, we will consider holding a seminar to take the discussion further.

In the absence on any advisor having any view on the perils of electronic advice provision, the Legal Services Commission are now tendering for a contract for a software provider to build:

an 'Expert system in debt and welfare benefits law'.

The aim of the project is to develop and build an expert system which will offer legal information and advice in debt and welfare benefits over the Internet for members of the public. The system may be expanded to cover other categories of law at a later date. The law reflected in the lines of enquiry will be that of England and Wales only. The system is intended to act as a self-help tool for legal information and advice.

For more details, have a look at this link

Anyone worried yet - because the more funds that are diverted to projects such as this, the less that is available for directly funding contracts.

The more access made available electronically, the less need there is for face to face advice.

Given the worries over Project Eagle (described in the Computanews article above), is it realistic for the LSC to be commissioning further electronic advice systems at a time when an existing project doesn't really seem to be fulfilling it's intended functions?

There would be no surprise if funding went to e-advice projects. Politicians do not seem to have a high opinion of advice services (if they did they would fund them)

Surely any e-system will be of limited usefulness. There are a long list of problems that require decision and judgements to solve:

1) Finding out what the problem is - clients know the symptoms not the cause of the problem.

2) Getting accurate information from client or benefit authority.

3) There is an implication that an e-system will be pitched at a level that an average person would be able to use. Not all clients are articulate or literate - these are often the people most in need of an advice service. How can an e-advice system pitch advice at a level the client can understand - how much explanation is needed.

4)What level of advice does the client need or want. Information - signposting - advocacy by phone or letter - representation in formal procedings.

5) Benefit decisions are not always correct. Some benefit authorities have a pityful standard of decision making. Even the average ones are not all that hot.

6) How can a "system" be "expert". Expertise comes from knowledge, skills and experience.

In the end an e-advice service would be useful to a small section of the poulation only (those with better than average articulacy) and probably as a reference tool by advisers.

The surprise here is not that the idea is new, it's that they may not have realised how much is already available.

To answer some of Derek's points,

If I was to get involved in an adviser v computer debate then I'd end up saying, I suspect, that a well designed computer system will be consistent and thorough in its information collection. It will not forget to ask questions, nor will it make incorrect assumptions and it should be able to determine what questions to ask or not ask according to circumstance. An advisor may well be able to focus quickly on the core issue(s) and information but may not choose to take a comprehensive view of the situation.

This covers MTBs in the regulated schemes. For NMTBs it is of course much more difficult to actually determine entitlement but it may well be that algorithms can be as good, or better, than advisors in estimating the probability of success for the generality of claims particularly in the indicator based schemes. They will not be as good where 'local' knowledge is important - "Try and see Mr. Jones - he's a soft touch for this condition" - although that kind of knowledge can be encapsulated.

There is a lot of experience and technology around that can help meet the different needs of users. It is often easier to provide a system that works in a minority language than to provide an adviser who speaks it. There are all sorts of ways to meet the needs of users in other ways, multi-media techniques for the literacy problems, avatars for sign-language etc.

Pitching the level of questions and explanation for different kinds of users is not new either. There are a lot of web sites which have had to do this for commercial reasons and, crudely, most make an assumption that the users of the technology need a minimum level of understanding to get to the site in the first place. Otherwise they will assume that there is an intermediary involved. It will also be true that no-one would attempt to produce a system that 100% of the population could use on their own. The question will be 'What's the target usage figure?'.

As for what level of advice does the client need or want.

I wrote a paper seceral years ago which broke electronic information down into a number of categories:

"Scatter Broadcast
For example, Posters, Advertisements.
Information believed to be of general interest or of importance to a smaller number of people. Probably not relevant to most readers / viewers and normally happened on by chance.

Leaflet
Provided by search effort of user.
This is the lowest level of information provision which requires the user to search for and find a document which may contain information of interest to him/her.


Target Broadcast
For example, list server, channel, but sent to those who express an interest.
The same level of information as leaflet which may contain variable amounts of information of interest or relevance to the user but which is 'pushed' to those who have previously expressed an interest in a particular topic.

Structured
Document assembly by dialogue.
The generation of a document making use of responses from the user indicating interest which selects from a boilerplate library elements which are of specific relevance..

Anonymous personal
Modelled advice - e.g. financial. No ID needed.
The provision of a service based upon a 'model' of a situation, for example, a social support query may be made by describing a situation of the user's family circumstances which will enable a calculation of entitlement to be made but without any requirement to provide personal identifiers.

Full personal
ID provided, allows use of data - a secure and digital signature.
The use of an identifier, smart cards, digital signature etc. allows for ‘change of circumstance’ and 'what if' scenarios using actual data already known about the user.

Interactive
Allows response and claims etc.
A fully secure identified connection allows responses and claims, for example, to be made completely electronically.

Personal structured
A document assembly technique making use of personal information which can be used to assemble individual documents such as wills.

Real time
Which allows instant interactive communication between government and citizen and may include, for example, on screen video links and document transfer.

Many of these may be enhanced by use of facilities such as automated translation, timed requests, etc."


I think that this kind of division is still relevant. No-one is going to expect the robot rep. to appear as an advocate at a tribunal but the human rep might very well have electronic information resources.

The calculations of benefit entitlement and amounts is a long proven given, even though it was widely considered impossible in the late 70s and early 80s when I started working in the area.

It doesn't need to 'expert' in the human sense, it needs to have a sound collection of rules in place and to be able to apply them correctly. The expertise comes in capturing and producing those rules which will not all be be lifted from legislation.

The more difficult area of this proposal will be debt advice where tactics and strategies come more into play and personal circumstances, which are harder to map, will become important.

I wouldn't agree that "an e-advice service would be useful to a small section of the poulation only (those with better than average articulacy) and probably as a reference tool by advisers" If it's a decent tool then it should be useable to a large section of the population, at least 80%, and it ought to be inappropriate for advisers who would probably need a different sort of tool with an interface that was appropriate for experts.

20% is one fifth of the population. that's a lot of people. it's not just about numbers. if that fifth is the most vulnerable and weakest part of the population, a special duty is owed by the statutory authorities to ensure that their interests have protection. the wrong IT 'solutions' can add to the problems, and software 'solutions' are not possible where humane judgement is needed. consider the DLA and Appeal Wizard partnership with IBM. different sectors of course. and the post-Moyna decisions. what percentage of the 'find it difficult to access traditional advice services' fall in the 80 % who would benefit? (just using the figures as 'for examples' rather than statements of statistical accuracy.)

FOI could send the system into overload, and i'm surprised if the LSC et al have recognized its positive potential for development.

i'd really like to know where the LSC is coming from on this, and what implications it has considered, and what it hasn't considered. there are some serious joined up thinking issues, as far as i can see. resourcing is just one. efficiency is another...

right now i'm going to sleep on it.

jj

I hope that they are sensible enough to realise that this type of solution can only meet some of the needs of some of the people. If it's not a pure cost cutting exercise the resources freed up using this system, where it is appropriate, can be directed to meeting more of the other needs.

I read Gareth's post with interest. Can't help thinking we are talking about different things. It's a common problem in IT systems that the system designers do not have sufficient knowledge of the subject matter and that those with the knowledge are unable to explain it in IT concepts.

I have to wonder if Gareth has much recent experience at the sharp end of delivering advice.

Using benefit calculators (and I use a Ferret one myself - and very good it is too) or series of questions on entitlement issues is at the very simple end of advice giving. It is why these systems can be and already are computerised.

I have to say that Gareth's concept of literacy does not match mine. I know several languages (particularly ones with multiple dialects) where clients are highly fluent yet have very poor abilities in the written language. I meet clients (all languages) frequently who are frightened to fill forms in at all let alone having any ability to understand technical bits and jargon.

Cannot see why any of these problems will be addressed in a computer system. If you have a lower than average articulacy or literacy then these will be a barrier to accessing or understanding a computer system.

It all comes down to one word "communication" I define this as "the transmission, reception and COMPREHENSION of information".

Computer systems can efficiently transmit and receive information. Its that last word - COMPREHENSION - where advisers come in. As an example I would challenge anyone to explain Non dependant deductions for HB & CTB - in a simple manner. I've been trying for years and still have difficulty explaining it to clients.

The dilemma in an advice system design will be that making it plain and simple enough to understand means that the complicated bits will have to be ommitted. If you leave the complicated bits in, it cannot be plain and simple.

I'm not worried about computer "advice" systems in themselves. If you are articulate and literate enough you will be able to use them. Unfortunately this excludes a large majority of my client base.

This may not be inconsistent with Gareth's 80% because I'm not saying
that a majority of the population are inarticulate or illiterate but that articulate and literate people do not usually seek or need advisors and are not in my client base.

What worries me is that government will ignore practicalities and assume that computer advice will reach everyone. They will surely be tempted to use the concept to save money.

Derek makes some important points here that are worth some responses (fight, fight).

I entirely agree with his common problem and it's one of the reasons why we think we can produce decent products, we do have people who can blend the two sets of skills. There is an increasing body of people who do understand both sides of the fence.

I'll hold up my hand to not having been a direct adviser for a lot of years but we do put in a lot of effort to trying to ensure that we understand what the issues from the sharp end are. I'm on the board of the local CAB, spend a lot of time talking to advisers, read forums like this, talk to policy makers etc. I think we probably make a good stab at being as informed as most second tier organisations can be. All of our profesional staff have been front line advisers at some point in their careers and it's pretty hard to forget what that was like.

I'll also agree that benefit calculators are at the straightforward, rather than simple, end of advice work - the information end.

where I'm going to disagree is his opinion that computer systems can't be made simple enough for *most* people to use succesfully.

Yes there are lots of cases where people who are pefectly competent and fluent are not literate in a language. This need not be a barrier. It's possible to have an audio component which reads out the question and to have an entry system which can be as simple as "press the square red button for No". Numeracy is more difficult, it's hard to collect figures of rent or income without that but it's also less of a problem, so researchers say.

The inarticulate may well be happier with a screen that a person who they may feel is metaphorically drumming their fingers impatiently while a computer will have infinite patience.

Making a computer dialogue understandable doesn't mean having to leave out the complicated bits; it means breaking those down into simpler and simpler sub-elements until thay are comprehensible. That's why I said that a system which could do this for a large number of people would be inappropriate for advisers who would find the dialogue for a one-time, untrained, inexpert user would drive them bonkers after the first couple of uses.

Gareth, one of the crucial aspects of any such project must be a clear understanding of what it is trying to achieve, and this is what I feel, fundamentally, these proposals do not have.

As an example, I use a quote you make in a post prior to this one:

"As for what level of advice does the client need or want.

I wrote a paper seceral years ago which broke electronic information down into a number of categories:"

Within a sentence, you have moved from asking "what level of advice" into breaking down "electronic information" - this is the crux of the matter for me, the different skills involved in providing someone with information and with offering someone advice. The two functions are not one and the same thing by any means, yet they are easily conflated within proposals such as those that the LSC are making with this scheme. I would agree that people can be guided, in a relatively sophisticated manner, to information that can help them better understand why they are in a situation, and helping explore possible avenues to improve the situation. I would agree that this is something that, increasingly, can be, and is being, done electronically.

What I don't agree with is that the LSC will be able to build an "expert system which offers legal information and advice in debt and welfare benefits", especially with the kind of restricted budget that will probably be made available. People in need of advice, often as the first stage, require reassurance that they can do something about their problems, before moving on to actually dealing with these problems. It's the commonly quoted example of a client with a housing eviction notice, that relates back to rent arrears, that relates back to housing benefit delays, that relates back to post going missing, etc etc etc. Unpicking these kind of queries are what an adviser does, and the LSC's own research has found that the people most in need of advice are also those who are most likely to have interlinked or overlaid problems, not something easily dealt with electronically by any stretch of the imagination.

Also, the legislation and caselaw that frames social security and tax credits is enormously complex and everchanging so any expert system would appear to be incredibly labour intensive in maintaining an adequate level of expert knowledge relating to the subject matter. Advice provision in welfare benefits is about far more than simply calculating someone's entitlement to any given benefit(s) (and anyway, if they need software to calculate that, they can get in touch with you ).

Finally, what has happened to Project Eagle lately? Where is the evaluation of the outcome of that project? Surely, there should be a public report that can usefully inform any of these discussions. I am deeply worried that the thinking behind this latest wheeze is to provide advice on the cheap, that the project is setting itself up to fail before it begins by taking on such a massive undertaking rather than, perhaps, focussing on a discrete area that may be more achievable. Can computers advise? Do the LSC listen? We shall see...

Paul points out the difference between information and advice and, I agree, the different skills and techniques that are needed for them. I haven't seen the details of this proposal yet but I would hope that it is clear about the scope of the project.

If we were to assume, falsely, that there was an infinite pot of cash and expertise for this project then we could decide whether it was possible to produce an expert system that could give advice.

When we produce training courses for advisers we are, in general, offering them a set of rules which they can follow. These sort of rules are of many sorts:

If someone says that they are being evicted ask them why
if they have rent arrears ask about HB delays
If they have HB delays then....


The essence of an expert system is that it tries to capture the knowledge of an expert and put it into these simple kinds of rules. The whole body of rules, taken together, and agreed by a group of experts (so that it's not one persons idiosyncratic approach) makes up something that could make the same judgments as if it was a human in the same situation.

This set of rules could be put down on paper, with a good index & TOC, and be usable. It's what the DMG tries to do to produce a group of decision making experts. On a computer the system does the looking up based on previous information.

Given the infinite pot there would be no technical reason why it couldn't be done. Given the real pot, there are considerably more uncertainties and doubts.

Legislation and caselaw is complex and everchanging but we have for almost 25 years been able to maintain systems which depend on it.

The question that the botomless pot encapsulates is simply, are there enough resources to meet the requirement and that, in turn, leads to; is the requirement well thought through and defined?

If the requirement is something predicated on providing early, simple help to prevent things becoming serious then that will be very different from trying to capture the expert knowledge of the Commissioners and the reps who appear before them.

But we already have an expert system - it's known as talking to more experienced advisors, in a manner such as is currently supported by the LSC's Specialist Support Pilot, or as provided byway of a website such as Rightsnet.

If there were more resources put into funding frontline caseworkers (not cases), and there were more resources put into publicising where clients can go to find information and receive advice on legal problems, and there was greater and more effective public legal education so that when things go awry, clients knew that they did have rights and these rights are enforceable, we may not have a situation whereby 1 in 5 people with a justiciable problem do absolutely nothing about it.

I definitely feel that electronic methods of information provision and education are worth exploring further, don't get me wrong - I just cannot escape the feeling that this project is the wrong way forward and the £££'s spent on it could be much much better placed being diverted to projects which are delivering real support, as in detailed in the 1st para. These proposals feel like an attempt to shoot the moon...

It's back to the bottomless pot, I suspect.

If there was plenty of money then maybe they would fund lots more advisers.

If there isn't, as we all suspect, then can you use technology to spread that scarce expertise more widely?

It may not be perfect, it probably won't be as as good as a decent adviser ... but, could it be better than nothing at all?

If the answer to that is yes, then how good does it need to be to be acceptable? In other words, is there a level of usefulness, that while better than nothing is unacceptable?

I have a suspicion that our views may be coloured by different geography. In Wales you may easily find that the nearest advice agency is a couple of hours away by public transport and there often isn't any public transport. London is very much better supplied. I'm not saying that services in London can meet the real demand but there's more chance for accessible services there than in Wales.

I share Paul's fears that the LSC may see 'expert systems' as a way of providing low grade advice on the cheap, but I also agree with Gareth's contention that IT systems can do useful things and should be pursued.

Since we have too many people chasing too few services - advice deserts and two week waiting time for appointments - it seems to me worth looking at ways of improving the range of services available.

Advice services tend to focus all their efforts at helping people who come through the door, and rightly so, but a lot happens before that and I would like to see more emphasis on provision of information to support people in dealing with problems, including information on where and when to go for help. That's why ASA set up www.advicenow.org.uk and why we have undertaken our recent consultation on legal education.

I see 'expert systems' at the information end of the spectrum. I know individuals at the LSC do see them delivering advice, but this surely is shooting at the moon - a Turing machine that can discuss debt??

It's noticeable that companies with vast resources like the banks and Microsoft haven't gone down this road in dealing with their queries, so what can the LSC hope to achieve with a more modest budget?

I hope they scope their project more modestly. If they can produce something useful on a particular topic for a specific client group, they will have done well. It seems to me the challenge is in identifying unmet needs and finding an area where technolgy can fill a gap. It isn't about replacing advice but providing an additional and complementary service.

(These are my own views and don't reflect ASA policy.)

i share paul's views on this one.
there is no bottomless pit, when it comes to public funding, and i think we must assume the default position that the money will be taken from elsewhere in the legal help budget, unless the LSC wants to leap in quickly to reassure us that we have this wrong, which it's welcome to do.

some of the other discussions on these forums give further cause for pause. i question the LSC's targeted users premise - who are the people who 'find it difficult to access traditional advice and legal services', and what exactly are the causes of the access difficulties?

if the LSC is not clear on the problems and their causes, how satisfied can it be that such a project holds the solutions and this is where investment should be made? the whole thing has a ring of the 'hard to reach groups' identified by the DWP - see the DWP partnership thread on the pension credit forum. i posited that the 'hard to reach groups' are created by the DWP's closure program and the real problem is hard to reach social security offices.

i accept that's something of a simplification, and not the whole of it, but i didn't invent the term 'advice desert', and imo, the loss of access to social security offices, AND a loss of front line access to legal services and advice on welfare rights and debt through funding cuts, will have a seriously negative impact on poverty and social injustice.

something else - on the incapacity forum there's a thread which makes very clear how the impact of computerised medical reports is opposed by advisors. there was also somewhere here a discussion touching on the secretive development of, incredibly, a DLA decision-making package, and appeal wizard.

a phrase that arose with the DSS computerisation was that of 'handling exceptions', at which it admits that it is poor. the term is of course a euphemism and a cover-all, and it covers up both the fact that human judgement is called for in some decisions which simply cannot be made by a computer program - there's a quality in wisdom that isn't arrived at by a linear process of if X then Y, and in some judgements, wisdom IS called for. The other concealment is the lack of training, knowledge and expertise - and levels of stress which mitigate against quality - all of which are connected to costs savings.

the faculty of human judgement is subtle. so is the law. computer decision-making is crude in comparison - this is why the computerisation of medical reports, no doubt an attempt to "capture the knowledge of an expert and put it into these simple kinds of rules" - as gareth explained, are a failure in all respects, except saving the government money, and improved legibility.

as advisers, we use subtle human judgements, as well as interviewing and listening skills, all the time with clients, especially with stressed and variously vulnerable clients, who need a lot of help to articulate their problem and give relevant information. we know that if we do not get full instructions, we can give bad, even harmful advice. there is not a computer substitute for these skills.



jj

After reading your thoughts jj, I am now picturing a system whereby clients go to a computer for some electronic advice about their benefit problem, before that advice computer liaises with the DWP computer that makes the review decision about the benefit itself....who needs humans anymore?

But more seriously, I do agree with your statement that "the loss of access to social security offices, AND a loss of front line access to legal services and advice on welfare rights and debt through funding cuts, will have a seriously negative impact on poverty and social injustice" - it seems odd to me that a government that purports to be committed to social justice and community cohesion continues to undermine some of it's better intentions and programmes by this lack of "joined-up thinking", to coin a phrase. To me, the buzzwords of "social exclusion" are best represented by the findings from Causes of Action (the LSRC research from last year) which found that 1 in 5 people with a justiciable problem do nothing at all about resolving that problem, leading to the conclusion that there is a:

"profound need for knowledge ... about obligations, rights, remedies, and procedures' for resolving justiciable problems"

The absence of knowledge that you have rights, the lack of understanding of what those rights are, and the lack of knowledge of how to enforce those rights will, in my opinion, inevitably lead to degrees of disengagement from wider society.

As the research notes, "the nature of justiciable problems requires that they should be of general concern, and that their prevention and resolution should be seen as a central part of efforts to tackle social exclusion...suggests that dedicated advice services should mirror more the needs and behaviour of those who wish to use them...highlights the importance of equipping those from whom people initially seek advice with the means to quickly and effectively refer them on to the most appropriate adviser, and the importance of accessible general advice services that act as formal gateways to the great array of advice and legal services...it suggests that public investment should come from across government"

I am concerned that we are in danger of falling into the old WIBBI style of discussion (Wouldn't it be better if).

Yes, I entirely agree that it would be better if everybody had easy access to expert advisers, government offices, proper support etc.

They don't; and this government, nor probably any other, is not going to spend the kind of money that that would require or change the organisational structures that they have got planned for benefits administration.

If you think that they can be persuaded to change their minds then by all means let's campaign for that but if you don't think that, then don't we have a duty to see what are the best services that can be provided for the people who need help.

There is a tendency in the advice world, as in the voluntary sector in general, for a kind of IMBYISM to take root (In my back yard...). By that I mean that efforts go into maintaining and developing services in the local area and community. That's entirely understandable as it's where their interests lie. I don't recall seeing any funding campaigns by local agencies to move funding from their areas to other parts of the country. It's also the reason why, for example, CABx used to follow the volunteers rather than the need. Someone would decide that CABx were a good thing and there were people with skills, experience, contacts and articulacy in the area able to do the work to start one. There were always people who needed the service, even in the most affluent areas, but it did lead to the development of service provisions that fitted badly onto a national map of needs. There were, and are, a lot of areas where those kinds of skills and vision didn't exist in the same way and nobody started the service that were needed.

If you aren't going to persuade the government to fund the roll out of services nationwide and if you can't persuade existing services to close themselves down in order to provide a more equitable spread of services (even if grossly insufficient) nationwide; what do you suggest is done?

If you look at the budget for this project (about £250,000) how would you spend it? Pay for about 6 advisers with their associated costs? Where would you choose to put them for maximum effect?

Doesn't it make some sense to try a project that might make better information available to a wide range of people, that might help stop some people from making the wrong decisions or help them make the right claim earlier? Aren't there lessons that will only be learned by trying these sorts of ideas out?

Even if you think that the most that can be done is to provide a pre-first line advice service, isn't that worthwhile?

Yes Gareth, it does make sense to try out different approaches, that's one of the commonly recognised benefits of funding voluntary sector agencies who do things differently.

However, it's just that those of us here in the voluntary sector usually have to jump through a hundred and one hoops in order to secure relatively modest funding that often is extremely time-limited (1-2 years commonly), that requires rigourous monitoring and evaluation, and that clearly defines at the outset (and for the duration of the project) what it intends to do, who it intends to reach out to, and what the effects or outcomes are intended to be.

Whereas this project (1) seems ill-defined in what it sets out to achieve, (2) does not demonstrate that any learning about similar projects, e.g. Project Eagle, has been taken on board, (3) is seemingly committing funds for up to a 5-year period in what looks like a strategic void from what is happening within the rest of the Legal Services Commission, and (4) it seems (from analysing what you and others write) that the idea of an electronic expert advice system is anyway, fundamentally, an unattainable goal - so we are left with an assisted information resource. Um...anyone used www.adviceguide.org.uk or www.advicenow.org.uk lately?

What about spending the £250,000 you mention on a public legal education project, so that people actually realise that they have rights and feel that they can do something about enforcing them? Or on a website that offers support to advice workers? Further, advice done on the cheap often results in bad, or worse still, simply incorrect advice being given, as jj notes. I agree with Martin that, if this is the amount of cash to be spent, then the LSC would be much better off designing a system around a specific issue, a specific community or locality, or a specific unmet need and evaluating if, and how, an electronic information system can meet any of these specific needs, complementing any other services already in place. I agree with you that advice provision in rural areas could well be bolstered by an electronic information system, and that there could be value in a electronic "pre-first line advice service" but I see nothing wrong with criticising projects that do not appear to move us any closer to realising either of these two aims.

Given the recent record of government-commissioned IT projects, perhaps we should just give EDS the £250k and then forget about it.....

I'm surprised that no other advisor has a view on this. Perhaps this is because advisers reflect the general public where a significant number of whom are still suspicious of, or at least cautious about "computers".

To try and get back to Paul's original questions - nothing in the discussion between Paul and Gareth changes my view.

1. There is nothing wrong with a new computer information/advice system. I welcome it.
2. It will not however reach all groups. It will be useful for the computer literate and if straightforward enough, for the average to above average articulate.
3. Few of the people in 2 above seek out advisers or get into sufficient problems to be referred to advisers.
4. It is a bit of a false comparison therfore to compare this computer system to what most advisers actually do.
5. Gareth says that a system is possible that would be accessable to anyone, whatever their difficulties. I agree with this in theory but I think I know my client base. The more I think of it the more barriers seem to arise.
6. To put it as simply as I can - to be accessible by my clients it would have to be extremely simple and to give accurate advice useful to the clients individual need, it would have to be very detailed(aka complex). Only a very, very, very sophisticated system could even begin to solve this paradox.
7. Whilst being theoretically possible, I do not think it is currently practical and something this sophisticated, currently affordable.
8. So I think that the system we will get, whatever is claimed for it, will not meet the groups in most need i.e. the client base of advisers (let alone those more needier still, who do not even get to an adviser)
9. The problem I have with this project is that I fear decision makers in government establishment will take a simplyfied view that if the sytem CAN be accessed by a large number of people then advisers are no longer needed and funding should stop.

I think I'd take issue with quite a lot of Derek's list but it does bear a strong resemblance to what people were saying 25 years ago about calculation systems:

It can't be done
If it can be done it will be too difficult to use
If it's not too difficult to use it will be too expensive
etc.

The one that's new to me, and that I would disagree with very strongly, is the contention that few computer literate people get into sufficient problems.

I'm obviously not getting the message over very eloquently.

There is a distinction between whether things can be done and whether they are currently practical or currently economic.

If that statement were untrue I would not be typing this into a keyboard, I would be dictating it direct into the computer. Like on STARTREK I could talk normally into the computer, hold a conversation with it and keyboards would be redundant. Similarly one of my clients could walk into an advice centre, talk to the computer (even though he does not know any terminology or which benefit is causing his problem), the system could discover his problem and provide a solution there and then.

Somehow Gareth, I do not think this is going to happen in the forseeable future.

This does not detract from any system that is practical and affordable. But don't confuse the possible with the practical.

Oh and your comment about what was being said 25 years ago reminded me of something a systems analyst said to me on my first day in my first job in computers over 30 years ago (yes 30).

He said (pointing at the room filling machine) "This is not a magic box. Unfortunately you will go through life meeting many people who act as if it is. People have a mental block and treat computers as things apart from the people who design them and run them. Computers go wrong but do not make mistakes. Mistakes are always the fault of a person. Computers cannot solve a problem it was not designed for"

How true.

I entirely agree with your systems analyst. That's why if you're looking at criterai for success in expert or rule-based systems the important ones tend to be getting the right expert rather than system.

If you'd like to talk rather than use a keyboard then you could look at Dragon Naturally Speaking, or one of the other similar systems, that millions of US users and hundreds of thousands of Europeans use to enter text and control computers.

At last - something we agree on - the right expert is the most important.

By the way I have tried voice recognition systems.

grey dont work merry welt

Flobba Dob, Weed.

Actually, the latest versions of them are extremely impressive, even when there isn't a specialist vocabulary involved, which is where they used to work best.

Still none that work in Welsh though, although there's been some very clever work on spell checking which was aften thought impossible because Welsh is highly mutated.

So if we agree it's worth trying something, and we want the LSC to scope it sensibly, what would we focus the system on? Apparently they are going to start with a 'proof of concept' pilot - and so need (I think) a discrete, easy, manageable topic that they can test with different client groups.

It needs to be largely procedural, without decisions based on discretion or fine interpretations of the law, and it needs to fill a gap - that is provide information where existing services are failing.

I think that presentation is key for these sorts of system - it would be doing a good service if it could step people through a complex process, say with a commentary, or break tasks down into manageable steps - Wizard style - or provide a visual route-map?

So what would be a suitable and very specific topic - from within Debt and Benefits?

This is the crux of the issue - for example, it would be, I would imagine, relatively straightforward to build a system that could take you through the procedures involved in challenging a decision to refuse you a benefit i.e. right to request review, timelimits, outcome, further appeal to tribunal, hearing, etc.

However, to move beyond these procedural aspects and delve into the substantive issues around benefit entitlement almost inevitably leads to questions of degree, to consideration of relevant case law, validity and weight of supporting evidence, etc. Not an easy conundrum to sort out.

So your electronic benefit calculator says that you should be entitled to the benefit, your local DWP computer tells you that you're not entitled, your electronic adviser tells you how to challenge the decision but who can explain the how and why you have been refused and who will help you begin to understand the grounds that you have been refused on and how to challenge them?

Got to dash now until next week so will give some more thought and eagerly await to see whether this issue has moved forward next week.

I'd start by making some bold assumptions.

If you want to produces something for general usage it makes sense to pick common issues.

It's more sensible to aim at something early in the process. That's when simple actions prevent complex problems.

Decide whether you want to deal with:

Entitlement
Claiming
Challenging
etc.

My guess, however, is that this being the LSC, they will be assuming a 'legal' issue rather than anything else.

I'd look at something based around regulatory law for clarity rather than areas of discretion or judgement. The latter can be done but determining weightings and probabilities should push the costs up if done properly.

I'd stick to 1 benefit or you couls end up in some circularity.

I'd try to focus on the end result rather than the reasoning process, unless the LSV see this as a technology pilot rather than an advice pilot.

Gareth

Interesting. Could you expand a little on the need to focus on the end result rather than the reasoning process?

And surely it is both a technology pilot and an advice pilot?

Martin

OK, I am using a smartphone so I"ll keep it short.

LSC have said they want an "expert system" not that they want a system to do particular job. Is it the job, regardless of programming tools, that they want or is it the ES experience they are interested in?