Lisp and Prolog appear in the European Commission's eGovernment Benchmark 2025

[triska]

Dear all,

the European Commission has recently published the eGovernment Benchmark 2025.

In a markedly more concrete statement than two years ago (see #2077), the Background Report now mentions among the Key success factors of the Good Practice example Grants4Companies on page 68:

Fast development due to standardized AI programming languages (Lisp, Prolog)

Grants4Companies uses Scryer Prolog extensively, and the report also links to the application's System Description.

Thank you all for making this possible, and enjoy!

All the best,
Markus

[hurufu]

Fast development due to standardized AI programming languages (Lisp, Prolog)

I think that marketing Prolog as an AI language actually served badly in the past, a lot of people think that it is DSL only good for symbolic AI and treat is as an embeddable black box just to be used from another language.

IMO Prolog is a good general-purpose programming language, and this notion shall be promoted too.

[hurufu]

I'm not sure if I agree with the first statement, but I haven't read the whole paper, so I don't know the details. I think it is detached from reality of huge real world datasets and impracticality to encode them using Prolog syntax – they are just too big. It seems that in most cases they use so called triplestores – and true they are connected to Prolog in some way, but there is a lot of tooling that actually enables people to work with them. I'm not convinced that it is feasible to encode such datasets using plain text files and Prolog syntax.

But Prolog shines as a programming language – because some very complex problems can be quite easily solved in it.

[haijinSk]

I agree, of course, and thank you... Yes, Prolog as a general-purpose programming language, and solving, at least some, very complex problems easily with Prolog. I know, it's implicit in "very complex problems [...] easily", but I wanted to add a "small and simple programming".

Edit: Also, I wanted to add the idea of the "knowledge representation language", the paper; then I deleted my comments.

[aarroyoc]

Another person here that agrees. I think Prolog is a good general programming language and we shouldn't restrict ourselves to categorize it as an AI language (especially now that the general understanding of the word is shifting towards other things). It won't be the first language that changed focus after creation. Java once pushed for web client execution via applets, now it's mostly used on backend servers where having a virtual machine is not that useful, but still.

[hurufu]

Quoting tech report from '93:

[...] To the commercial computing community, the terms “Prolog” and
“logic programming” are at best perceived as useful in an academic or research setting,
but not useful for industry. This image is not based on any rational deduction. Changing
the image requires both marketing and application development.

Progress exists with the rise of neuro-symbolic and/or explainable AI systems, but not in application development space. Prolog still needs marketing, applications and tooling. For example compare with quite young¹ Flix language and the amount of tooling and marketing.

Footnotes

1. Started circa 2014 ↩

[triska]

So it started 2 years earlier than Scryer, it will be interesting to see how much Scryer improves in the next 2 years, and compare it then to what we have today! Personally, I hope for very significant developments.

[triska]

One of us wrote several interesting and insightful comments here and then unfortunately deleted them!

Personally, I found the comments great contributions and would have appreciated if they had remained also for others to see and discuss further.

If possible, please consider reposting them. Thank you a lot!

[aarroyoc]

One good aspect of Prolog, right now, is that nobody "owns" the language. Or put it differently, many people own it at the same time. There's not going to be a company that suddenly makes changes and leaves everyone upset. No changes to maximize profit. This is not to reject companies, but I always felt that community approaches are more durable (probably more "antifragile" in N. Taleb terms).

[hurufu]

Yes it is a big plus and it allows organic evolution and growth, but also because nobody "owns" the language we have over 9000 different implementations and human resources are vastly diluted.

[UWN]

we have over 9000 different implementations

16

[triska]

We also already have roughly 3 implementations that can parse Prolog! It's a start.

[hurufu]

In the past I've decided that I want to package all Prolog implementations (and dialects if they use Horn clauses) for Arch Linux. It is still an ongoing endevour and I have counted much more :)