Inclusion of CoNext for computer networking, especially as a non-default option.

[sgrlaf]

Currently, the only conferences listed for networking are Sigcomm and NSDI.
This is to request addition of CoNext as a non-default option.

To put this in context,
--OS lists Eurosys, Usenix ATC and FAST as non-default options in addition to SOSP and OSDI
—PL lists ICFP and OOPSLA as non-default options in addition to POPL and PLDI.
-Architecture lists MICRO as default in addition to ASPLOS and ISCA, and lists HPCA as non-default.

CoNext is a solid conference and a clear #3 in networking, and at least on par with Eurosys/FAST/OOPSLA.
It also seems quite harsh to list only two conferences for networking, especially brutally competitive ones such as Sigcomm and NSDI.

I noticed that a request had been sent earlier: #6305
I understand CoNext not being included in the default rankings, but I do believe there is a strong case to at least include it as a non-default option.

Sanjay Rao
Professor
Purdue ECE

[waehlisch]

+1 This would be a fair compromise.

[aruna338]

+1 for including CoNext as a non-default option. CoNext has long been recognized as a premier venue for networking research, it features rigorous review standards and accepts top quality papers. Adding it would help represent the diversity and depth of the networking field, similar to how Eurosys or FAST complement OSDI/SOSP in systems. Between SIGCOMM and NSDI, around 150 papers are accepted each year. There are more researchers in the networking community producing top quality work beyond these 150 papers and CoNEXT is the natural next venue for strong networking research. Including the conference in CS ranking would help ensure that active researchers, especially junior faculty, will receive appropriate recognition.

Aruna Balasubramanian
Associate Professor, Stony Brook University (also on the Steering Committee and past program chair for CoNEXT)

[Rajrup]

Yes, I second the proposal to add CoNEXT as a non-default option in networking. CoNEXT community has grown over the years. The community seems to welcome newer ideas in networking, which are sometimes overlooked at premier venues such as SIGCOMM and NSDI.

Rajrup Ghosh
PhD Candidate, University of Southern California

[dongsuh]

I agree with Sanjay and Aruna. Great argument and what many of us would like to see.

Dongsu Han
Professor, KAIST (past program chair for CoNEXT)

[vyassekar]

Agree with this and perhaps this is long overdue.

Having chaired SIGCOMM, NSDI, and Conext, the quality of papers and selectivity is comparable.

Vyas Sekar, CMU

[psteenkiste]

CoNEXT definitely should be added to the Tist. together with Sigcomm and NSDI it is one of the three top networking conferences.

Peter Steenkiste, CMU

[emeryberger]

Some data to add context for this discussion:

Currently listed conferences:

SIGCOMM:

• rated A* by CORE
• 21% acceptance rate over last 5 years, ~350 papers submitted in 2024 (https://csconferences.org/#SIGCOMM)
• publication impact per Google:

Publication	h5-index	h5-median
ACM SIGCOMM Conference	64	103

NSDI

• unrated by CORE
• 17% acceptance rate over last 5 years, ~650 papers submitted in 2025 (https://csconferences.org/#NSDI)
• publication impact per Google (number 8 in "computer systems": https://scholar.google.com/citations?view_op=top_venues&hl=en&vq=eng_computingsystems)

Publication	h5-index	h5-median
USENIX Symposium on Networked Systems Design and Implementation	71	99

Proposed optional conference:

CoNext

• rated A by CORE
• 121 papers submitted in 2025
• publication impact per Google:

Publication	h5-index	h5-median
ACM Conference on Emerging Network Experiment and Technology (CoNEXT)	29	42

[GianniAntichi]

I think CoNEXT shall be added in the list as non-default option. This would be just fair.

Gianni Antichi, Polimi

[aruna338]

Emery,

There is some discrepancy in the number of submissions that are quoted above. For CoNEXT 2025 which has two deadlines—a cycle 1 deadline in December 2024 and a cycle 2 deadline in June 2025—256 papers were submitted. In fact, over the years there is an increasing trend in paper submissions. The acceptance rates have also been consistently around 20% for the past 20 years (from 2005). The submission and acceptance statistics can be found here https://docs.google.com/presentation/d/14I707yl1X4NhHt8iDd7UF9JAz3UxzMs5-DeotgXCKU0/edit?slide=id.p#slide=id.p

CoNEXT is sponsored by ACM SIGCOMM. Flagship conferences of ACM SIGs are usually considered core venues in their respective fields. It is also not totally clear if H-indices or citations are the most relevant metric since if you look at those metrics, then the top ranked venues in networking are journals that are not considered relevant venues by the networking community (and are certainly not in the CS ranking)

One idea here is to do an anonymous poll across the SIGCOMM and Usenix networking community to find out their perception of CoNEXT. Would that help in making this decision?

[sgrlaf]

• I second the notion that h5-indices shouldn't be the only metric. It isn't clear CSRanking is using h5-indices to base all its decisions either (e.g., Infocom has a h5-index of 80, but hasn't been included either). I suspect also the h5-index is dependent on the number of papers published at a venue as well.

• There seem two questions to consider: (i) is it consistent to only include only two conferences for networking (across default/non-default) when other communities have more? and (ii) if we agree there is a case to add non-default options to networking, what are the natural venues?

On (i), I think the bar to get a paper in core networking "count" as per CSRanking is quite high relative to other communities
right now (assuming both default and non-default options are counted), and there is a good case for some non-default options to be added. On (ii), I suspect most folks who are active in Sigcomm/NSDI see CoNext as the natural other option for a core networking paper.