Cancel ISCA!
Why Cancel ISCA
This has been building for years, but the death of
Huixiang Chen shows just how bad things have gotten.
Timeline of ISCA'19 Alleged Peer-review Fraud / grad-student suicide
See here for a full timeline of the
Huixiang Chen incident and the lack of effective response by ISCA,
SIGARCH, IEEE or ACM
General Background on ISCA
The ACM/IEEE
International Symposium on Computer Architecture is an annual
academic conference on the low-level implementation and design of computer
systems. It has been around since 1973.
The conference is considered by some to be the top conference in computer
architecture (in the computing fields
publication in conferences
is generally considered more high-profile than publishing in journals).
Because of its alleged high profile,
there is a lot of cut-throat competition into getting papers accepted
at ISCA.
Researchers who want tenure want to have papers there, and PhD students
are put under immense pressure to publish there.
ISCA Paper Quality
Over the years there have been rumblings about the paper quality at ISCA.
The methodology of the papers often seems weak, with a high reliance
on simulators (rather than real-world results).
We have started investigating the
methodology found in past ISCA papers.
There is no artifact submission for ISCA, meaning researchers
don't have to submit the source code and raw data. Often if you request
to see the data or simulators used, you are denied.
This means most of the work published at ISCA is essentially non-repeatable.
Due to the close-knit manner of the community there have been
questions about paper peer-review. Papers are double-blind at ISCA,
meaning in theory the author doesn't know who is reviewing the paper, and
the reviewers aren't supposed to know who submitted the paper.
In the relatively small computer architecture community it is hard
to keep such secrets and identities are often known or easy to find out.
It is common for a large number of Program Committee papers to get
accepted at the expense of outsiders, a
longer investigation of this issue can
be found here.
Recent revelations have come out about a student's suicide due to
methodology and peer-review concerns. The student had in his possession
unredacted reviews from the 2019 ISCA conference (as well as the related
HPCA conference) showing that a compromise of the double-blind review
occurred. The various ISCA organizers have done nothing to address this.
See here for a full timeline on the incident.
What can I do?
See here
Back to the main webpage