TIMES as open source

154 views
Skip to first unread message

Robbie Morrison

unread,
Feb 14, 2023, 12:21:38 PM2/14/23
to openmod list

Hi all

I am trying to find out which TIMES models are open source — more specifically:

  • the core TIMES codebase
  • TIMES-UK
  • JRC-EU-TIMES

The core TIMES codebase is located on GitHub but with some likely trivial license notice implementation errors that prevents GitHub (and doubtless other tooling) from identifying the applied license:

That repo is apparently only used for archival purposes and there is no evidence of development activity — not that that prevents a codebase from being described as open source.

To my knowledge, the TIMES-UK model was discussed as being released as open source for some years now — but nothing ever happened in that regard?

The JRC-EU-TIMES is supposed to be open source according to this undated web announcement (with broken links also):

And there is some further information on the JRC portal:

But I could not find a public repo anywhere, nor any information as to which license applies.

Can anyone shed any light on these various leads and riddles?

TIA, Robbie 
-- 
Robbie Morrison
Address: Schillerstrasse 85, 10627 Berlin, Germany
Phone: +49.30.612-87617

Robbie Morrison

unread,
Feb 14, 2023, 1:12:07 PM2/14/23
to openmod-i...@googlegroups.com

Hi again

Three more leads (thanks to an off‑list response).

Something here on the JRC EU TIMES model:

This looks like a collection of spreadsheet files (TIMES is written in GAMS as I understand it) with no license present — therefore not remotely open source.

There is an ON-TIMES model under a Nordic Energy Research account:

That is licensed under the long deprecated ODbL‑1.0 data license — so again not open source.  And again what looks like a collection of spreadsheet files.

And the TIMES-Ireland model is here:

Under a Creative Commons CC‑BY‑4.0 data/content license and again a collection of spreadsheet files — once again not open source.

The ETSAP programme, which resides under the IEA umbrella, describes the TIMES model generator in some detail, but not its legal status:

For background, all common public licenses, both current and obsolete, are listed here by the Linux Foundation:

Also, that the term "open source data" is meaningless.  Either "open source software" or "open data".  These are long established definitions.

It would appear that dumping a set of XLS files on GitHub, perhaps with a suitable data license or perhaps not, constitutes publication and transparency?  If that is indeed the case, that certainly falls well short of the notion of open modeling as advocated within this community.

What gave rise to my interest in TIMES is that Göke et al (2023:6) (who posted earlier today, see here) were led to believe that TIMES is open source and that it "has become the standard for new models".  Now that may well be the case, but I am having great difficulty finding any evidence for that position.

Repeatability, enabled both technically and legally, is surely the cornerstone for public policy analysis in this context.

with best wishes, Robbie

--
You received this message because you are subscribed to the Google Groups "openmod initiative" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openmod-initiat...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/openmod-initiative/9061d330-c941-bf4e-7973-25302577dbf6%40posteo.de.

Robbie Morrison

unread,
Feb 14, 2023, 2:07:20 PM2/14/23
to openmod-i...@googlegroups.com

Hi again

Apparently, if you want to use the official version of TIMES, you will need to enter a bilateral contract with the Technical University of Denmark.  You obtain a license to run the software, of course, but the remaining terms are restrictive and certainly not remotely in accordance with the Open Source Definition — the accepted touchstone for evaluating open source licenses.  I have a copy of that contract and will try and confirm its currency and veracity tomorrow.

If anyone can shed more light on these questions, please contact me off‑list or reply on‑list.

with best wishes, Robbie

Robbie Morrison

unread,
Feb 14, 2023, 2:28:06 PM2/14/23
to openmod-i...@googlegroups.com

Hello once again

Here is the full sentence from Göke et al (2023:6) — which I misread:

Since Keepin's critique of the IIASA model, which was only possible because he worked at IIASA himself, influential models like TIMES have been made publicly available and open-source has become the standard for new models.

So that sentence does not strictly say that TIMES is open source, but rather just publicly available. And by implication falls below the standard set by new models.  I interpreted the "and" in the wrong context — that said, the hyphen should probably not be present in "open‑source".

To note that some important legacy models way well be released as open source in due course — so these process can be undertaken appropriately.

my apologies to the authors, Robbie

Mark Howells

unread,
Feb 15, 2023, 2:30:10 AM2/15/23
to openmod-i...@googlegroups.com
Dear Robbie (and all)

I wonder (out loud) if it would not be good distinguish between: (1) environment/language; (2) solver; (3a) GUI interface (3b) Command-line interface (3b) ; (4) model-generator; (5) model; (6a) case (and (6b) scenarios)? Or similar, when referring to a 'model'. (Where 'case' and 'scenarios' are generally defined by external input data; (5) is often defined by the modeller) That may help the discussion a little especially with folk on the 'outside'.  When someone says 'open' they seem to refer to (4), (5) or (6a&b) - while the newb - or even the purist - may think its (1)-(6).  Not having the elements backed out allows for easy (and sometimes deliberate) confusion. The data in the Irish TIMES model that is in a set of open-licenced spreadsheet files may well define (5), and probably (6). So, in a sense that may qualify - if model is defined by (5 & 6), no? (Unless a spreadsheet file is in an of itself is not legally 'open'. Which may just be an innocent mistake on the part of the modellers for choosing xlsx over ods.) In any case, I think that we should be welcoming to folk who increasingly develop open (and better) elements 1-6?

Best, Mark
From: openmod-i...@googlegroups.com <openmod-i...@googlegroups.com> on behalf of Robbie Morrison <robbie....@posteo.de>
Sent: 14 February 2023 19:28
To: openmod-i...@googlegroups.com <openmod-i...@googlegroups.com>
Subject: Re: [openmod-initiative] TIMES as open source
 

Mark Howells

unread,
Feb 15, 2023, 2:32:40 AM2/15/23
to openmod-i...@googlegroups.com
Just to add with,  U4RIA, I think that we really should start call for the goal to report analytical and organisational (i.e. stakeholder engagement) workflows to those elements. Best, Mark
https://www.researchsquare.com/article/rs-311311/v1

From: Mark Howells <M.I.H...@lboro.ac.uk>
Sent: 15 February 2023 07:29

Robbie Morrison

unread,
Feb 15, 2023, 5:42:43 AM2/15/23
to openmod-i...@googlegroups.com

Hi Mark, all


Good points on definitions.  I have now moved to off‑list discussions with the TIMES folk and hopefully some licensing fixes and further clarifications will result.


I must admit I was entirely confused by statements like "The JRC-EU-TIMES model now open source!". (see URL given earlier)


It appears, on reflection, that there should be a greater distinction between the TIMES Model Generator and a TIMES model and the set of datasets supporting the latter.  Only the TIMES Model Generator is open source under GPL‑3.0‑or‑later and I am currently working through some niggles regarding the implementation of license notices and exceptions within that repo.


I also suggested to the TIMES folk that they move to Software Heritage SWHIDs, these being unique and persistent IDs (relying on a Merkle DAG) that can identify specific historical codebases, snapshots, releases, files, and even sections of source code by line number:

We, as a community, should make more use of this excellent and cool infrastructure too.


And also for clarification, the Oslo‑based Institute for Energy Technology (IFE) replaced the Technical University of Denmark as the "operating agent"  for the TIMES Model Generator about three years ago.  So some of my earlier comments in that regard were misplaced.


Finally, just to reflect from the sidelines that I am very happy that established in‑house consortium models are moving towards open analysis.  But also to note that there is really no halfway house and that that journey needs to run to completion at some point — and particularly if accuracy, transparency, repeatability, and contestability are also sought in the public interest.


with best wishes, Robbie

Robbie Morrison

unread,
Feb 19, 2023, 4:46:59 AM2/19/23
to openmod-i...@googlegroups.com

Hello all


My discussions with the TIMES community proved entirely fruitful. And I just posted my findings to the openmod forum, rather than here, in the interests of visibility and persistence:

The TIMES ecosystem is technically open source and can be described as such without issue.


The overall project is clearly moving towards more public development and I was encouraged by the TIMES community to post an issue on GitHub regarding the license notification bug I mentioned earlier. Briefly, GitHub was not detecting the required license properly and fix was merely to shuffle some file names. GitHub will identify the correct license with the next release of the TIMES Model Generator, namely GPL‑3.0‑or‑later.


There is an important social component to open source development, which is why I am being a bit circumspect regarding full openness. But I believe that TIMES team genuinely wishes to address that matter going forward.


Just for the record, the earlier restrictive bilateral contract for the TIMES Model Generator, with the Technical University of Denmark (DTU) as the counter‑party, can be found here:

Those conditions were replaced by the GPL license in full in January 2020. Indeed, any of the GPL licenses will automatically withdraw when additional contractual terms are imposed (the so‑called infamous "liberty or death" clause) so any kind of dual licensing in this context was not legally feasible.


Worth noting too that the model interface architecture long used by the TIMES ecosystem is similar to initiatives underway in other open projects, including the agent‑based FAME/AMIRIS project and more conventional PyPSA‑Earth/PyPSA model, to pick just two.


My thanks to the several people who offered information to me off‑list. That information was very helpful and much appreciated.


If there are errors or improvements to be made to the openmod forum posting listed above, please let me know.


All in all, a very good — and I must admit, somewhat unanticipated — outcome. Indeed it was a complete pleasure to interact with the TIMES community.


with best wishes, Robbie

Reply all
Reply to author
Forward
0 new messages