Skip to content
Eddie Kohler edited this page Jan 1, 2016 · 4 revisions

HotCRP TODO

Capabilities

  • Use capabilities for accept/decline review links
  • Right now “Decline review” requires confirmation; it would be better to make it HAPPEN when you click the link and then be undoable

Comments

  • Allow attachments on comments and reviews
  • Comment notification emails should include opt-out links

Paper options

  • Add multiple-checkboxes option type
  • Auto-convert numeric <=> text when setting changes
  • Box for chair notes (maybe a comment?)
  • Add options set by administrators?
  • Use a separate Delete button rather than form position

Paper editing

  • Add a “cancel” button on paper submission
  • Submission history

Schema

  • Make sure .htaccess files work with Require as well as Deny

Visual appearance

  • New logo: Paper with flaming review bubble?
  • Links to home and (for chairs) settings

Paper strip

  • "Follow" rather than "Email notification"
  • "Follow" is a button?
  • Is there a good "Edit" image rather than button?
  • "Edit" button for paper strip rather than folds
  • "(shown only to PC reviewers)" etc. visibility should be bar-separated, not parenthesized.
  • Comment edit links shouldn't reload the page
  • "All reviews" links should say "Main"
  • Maybe a separate little area with links for "main", "edit", "review", "assign", on left-hand side
  • Review ratings appear in review region, above review itself (don't overload the review region)
  • "None" for Discussion lead/Shepherd/Paper manager fold to nothing

Paper list

  • If anchor is selected, highlight that row
  • Edit pads have arrows on them, not dots

Old TODO

  • When reviewers update their reviews, all we get is an email that the review was updated. But there's no indication how it changed: did the reviewer update the text? Did they raise/lower any score? Ideally, we'd like to see what has changed, but HotCRP doesn't have versioning. Versioning would be a lot of work, so perhaps a quicker/simpler solution would be this: upon a review change, perhaps you can format the old and new review (plus scores) as plain text, and include a diff(1) output of the old vs. new reviews along with the email that notifies the chairs of the review change? [Erez Zadok/Dan Tsafrir]

  • http://harvesthq.github.com/chosen/ [Tom Limoncelli]

  • (1) It would be nice to be able to tag papers with their session number and session chair. However, tag:session#1 tag:sessionchairbalachanderkrishnamurthy somehow does't do it for me, partly because these two could get out of sync, and partly because I would like to be able to use "send mail" in a way that gets mail to the right session chairs.

    What I think I would ideally like is to be able to assign a session number to an accepted paper (manually only) -- maybe "tag:session#1" is the right way to do that -- and then be able to assign a session chair (from the PC members and/or external reviewers, in case not enough PC members are willing) to all papers with a given tag:session value -- something like assigning a shepherd.

    (2) Then I would like to be able to send an email to the session chair that includes the name and email address of the author(s) in her session, and also send the authors an email with the name and address of their session chair. In "Send Mail" I think this requires a way to select To: == session chairs (analogous to To: Shepherds), and also new keywords %SESSIONCHAIRNAME% and %SESSIONCHAIREMAIL%. Oh, and maybe %SESSIONNUMBER%.

    And %SESSIONTITLE%, for fun, but that would require a way to set up that mapping, and this probably means a new screen somewhere.

    (Extra credit if you adopt the suggestion that John made, a while back, to optionally send a single message with multiple recipients; for example, if we want to tell the 10 authors of one paper that "one of you should contact the session chair" and (a) not end up sending 10 emails to the chair, and (b) make sure that they are all CCed on whatever replies get sent.)

    (3) Then I would like to be able to download a spreadsheet showing papers, session numbers, session chairs (names and emails), titles, and authors (names and emails) -- I had to do this myself, somewhat by hand, and it was tedious.

    (4) Then I would like a way for the authors of an accepted paper to mark one of themselves as the speaker for the paper (sort of like being the contact author, except that probably it should allow no more than one). This turned out to be tedious, too -- it involved writing scripts to generate emails and then following up by updating a spreadsheet by hand.

    Extra credit: if I can send mail to "authors of papers with no designated speaker" so that we can automatically hound these miscreants.

    (5) And, I guess, the spreadsheet from #3 should have a type field that indicates which author is designated as the speaker.

    (6) Extra-extra credit: a way to assign a paper-within-session order, and then a way to format the entire program, with session titles, session chairs, papers in the right order, author names and affiliations, and the speaker's name in bold. [Jeff Mogul]

  • Two questions we got constantly were ... "how do I get back to the main menu?" (click the name of the conference in the upper left corner). Both of these indicate areas of improvement for the GUI. [Tony Limoncelli]

  • Add a "bulk update" that shows the papers in a spreadsheet-like grid with number, name, shepherd, accept status, submission type, and so on. Let the user edit any of the data in the grid. [Tony Limoncelli]

  • We have to collect information for the program such as "name as to appear in the program". It would be nice to have a facility to collect all that information AND be able to collect whether the author has approved/acknowledge it to be correct. (If we could do this for ITs and Gurus, this could produce the entire program) [Tony Limoncelli]

  • Feature 1: Multi-paper discussion. A few times per conference, I want to write on a review "This paper is better/worse than paper X because ..." or similar. The problem is that such comments are hard to do without risking anonymity. When I enter a comment for a paper, it would be great to be able to, say, enter the number of another paper(s) and treat this comment as conflicting to people for whom those papers are conflicting. [Mike Dahlin]

  • Feature 2: Cluster-based assignment. Walfish and I were discussing how useful it is to make sure that there is good overlap of reviewers across papers on the same topic. The thought was that it might not be too difficult to make the automatic review assignment tools identify clusters of likely-related papers (based on co-citation, abstract text, keywords, body text, ...) and to help the chair get reviewer overlaps within clusters. [Mike Dahlin]

  • In any case, it occurred to me that it would be very useful to have a semi-analogous feature that allows us to specify something like

    Assign at least [N] common reviewers to papers [X] and [Y]

    This seems to come up fairly often -- someone realizes that two papers cover very similar problems and/or approaches, but either they don't feel completely comfortable being the only reviewer who has read both papers, or they is us, and we want to make sure that some PC member has done a true review of both papers.

    I assume that this kind of request can overconstrain the assigner, but it seems like the kind of thing that might rank fairly high in the prioritization of assigning reviews (e.g., perhaps somewhat more important than minor differences in review preferences or workload.)

    In any case, I think we would want to do these overlapping assignments as early as possible in the process, as opposed to the seemingly usual practice of deciding during the PC meeting that "Eddie should read paper #97 during lunch". So if we don't get automated support from HotCRP, we will try to do it by hand. [Jeff Mogul]

  • Can we save the review forms from this offering of the WQE for use in the next offering of the WQE? Since the WQE is not an actual conference, the standard conference-style review forms are less useful, so it would be useful to keep what we develop for future offerings. [Amit Sahai]

  • Viewable history of review scores (and review texts?) [Yoshi Kohno]

  • Cookie Stealing Vulnerability [Aditya K Sood]

  • The download list is limited (what if I want authors AND scores at the same time). Doesn't it seem cleaner to use the existing diplay options to select fields, then have download > "list as tabbed text" that gets whatever you've selected? It would be also good to work this in to the chair help page for someone who reads the documentation. [John Heidemann]

  • HotCRP Feature Request: PDF Anonymization [Matt Johnson]

  • Online TPC meeting support [Richard Mortier]

  • PC members can use mail tool to send mail to authors

  • Add "Full author info" checkbox to manual assignment page [Jeff Mogul]

  • Do people who've submitted only abstracts get email in the rebuttal period?

  • Show PC when authors can see reviews.

  • It would have been useful to have a "mail reviewer" link on the reviews page for communicating with individual reviewers about specific papers. [Stephanie Weirich]

  • a way to see the distribution of topics for submitted/selected/accepted papers. I computed this information manually, but it would be nice if the system could automatically display the info. [Stephanie Weirich]

  • a way to list the accepted papers w/ authors suitable for posting on the conference website. Did I miss how to do this? [Stephanie Weirich]

  • it would be really nice if email replies to comments became comments. yes, there is an auth/anti-spam issue. but you know the emails of the reviewer set.

    all real noc ticketing systems do this. it's a users' expectation that, if i receive an email and respond to it, the response will have the same level of distribution as the message to which one is responding. [Randy Bush]

  • It would be nice if the To: choice in the Send Mail function allowed me to select "Reviewers with no preferences" (or perhaps "PC members with no preferences"). [Jeff Mogul]

  • I see your point; possibly this means that the per-paper screen that the reviewer sees should have an "I have a conflict" button that they can push, but only the chair can un-push it. [Jeff Mogul]

  • On the Conference Setting/Submissions Options page, it's a bit tedious to type or cut-n-paste all of the topics. Seems like a way to bulk-upload this list would be useful to future PC chairs. [Jeff Mogul]

  • At recent PC meetings that I have attended, it has been hard for the PC chairs to keep track of the amount of time spent discussing each paper (since they are juggling lots of other tasks).

    I realize that this could be a bit tricky to get right, but perrhaps HotCRP could have a "discussion timer" function, sort of a per-paper stopwatch.

    Probably there should be two values: the time spent in the current discussion, and the total time spent discussing the paper (since PC meetings generally have loops).

    The PC chairs would see a Start-/Stop- timer button.

    Possibly this timing info should be hidden from the conflicted reviewers.

    Additional possibilities would be to do fancy things such as estimating how much time is remaining to discuss the average paper, or how late the PC dinner will be if we keep spending so many minutes per paper, etc. [Jeff Mogul]

  • It would be great to be able to assign reviews manually, but then ask the system to look for swaps that would make both reviewers happier according to their preferences. I'd want to be able to confirm one swap at a time. [Andrew Myers]

  • Visually or in search differentiate external review scores from PC review scores [Casey Henderson/Adam Moskowitz].

  • When external reviewers submit their reviews, it might be a good idea to inform the requesting PC member that the review has been submitted. Perhaps this could be an option available in the settings panel ? [Moses Charikar]

  • More than once, PC members asked if they could upload pdf files they had received as reviews. In one case, I converted the pdf to text, but in general, it would be useful to have a pdf upload option for reviews. I agree that it interferes with viewing everything on one screen, but in some cases (with a lot of math) pdf is very handy. If this is supported, it would also be good to send such pdf reviews as attachments (if they are meant to be seen by authors) when reviews are sent out, or be available to authors online when the reviews are made visible. [Moses Charikar]

  • We didn't do a full fledged rebuttal phase, but I contacted authors on a case by case basis, typically forwarding snippets of reviews and requesting a response. I got back email responses which I posted as a comment. It would be convenient if the system had a mechanism to allow authors to respond on the conference server (without having all reviews visible to them) and have the response appear as a comment automatically. A couple more related things: a. It would be convenient to allow PC members to initiate such author communication (in an anonymous fashion). As PC chair, all communication to authors was channeled through me - such a feature would take some of the load off. b. Authors sometimes responded with pdf files, so it would be convenient to allow pdf uploads in author responses. [Moses Charikar]

  • As PC chair, I used the action log quite frequently. It occurred to me that it might have been convenient for PC members to have such a list too

    • restricted to papers that they were assigned to, or have commented on ... basically all papers that they would receive email notification about. This would allow them to scan through all the new things that have happened since the last time they logged in. [Moses Charikar]
  • In several cases, we compared multiple papers. To facilitate this, I added a common tag to a group of papers and also created a dummy paper to hold the comments relevant to that group. It would be nice if the system supported such a discussion. One idea might be to allow comments to have tags, and then the comment shows up for all papers that have the matching tag. Alternately, a comment could be associated with multiple papers and this could be specified when the comment is entered (this will be hard to do if the group of relevant papers is large.) [Moses Charikar]

  • One PC member said "The server lost the name of my external reviewer that I entered in the name field (on two occasions; see, e.g., #21), and rudely addressed the person by their email address. :) Now it does not allow me to change "[no name]" in the name field." [Moses Charikar]

  • I was going to put out the abstracts for accepted papers together with the accepted papers list. In the current system, is there a way for authors to update the text abstract they submitted ? I suppose not, because they could probably change their submission too if we opened up the system for edits. [Moses Charikar]

  • Related to the previous point, I may have to collect tex files containing titles and abstracts from authors. I know the system allows collection of final versions (and archives original submissions). Would there be a way for the authors to submit additional files too ? [Moses Charikar]

    1. It would be helpful to store information about reviews outstanding by reviewing round eg paper n has 1/3 for the first round and 4/5 for the second. In fact it would also be great to be able to look at paper rankings by round to see how things changed. I would really like stats to see the value of multiple reviewing rounds - how much first round reviews predict the eventual outcome etc. [Rebecca Isaacs]
    1. Can you list both authors and institutions in the conflicts page? And for that matter it would be helpful (to the chair at least) if the source of conflict was listed, eg requested by author (or PC) vs matching words in author list, etc. [Rebecca Isaacs]
    1. It would be nice to have multiple review forms, in particular to have a different form for externals. [Rebecca Isaacs]
    1. Have you considered functionality so the automatic review assignment can take into account nominated reviewers? Eg so 1st round reviewers can suggest 2nd round people through the system and the chair doesn't have to do reviewing rearrangements manually. [Rebecca Isaacs]
  • i like the fact that i can see - just based on simple averaging - how harsh or liberal a reviewer is. however in defense of harsh reviewers who may have obtained genuinely crummy papers and liberal ones who might have recd instant-accepts, what might be useful is once all or most of reviews are in, looking at the deviation of the 'harsh' reviewer from other reviews of the papers he/she reviewed, and likewise for 'liberal reviewer and recompute a balanced score. as long as this is available just before the PC meeting then during the PC meeting we would know who is a softie and who is trying to 'kill' papers. am not sure if this feature exists now. [Balachander Krishnamurthy]

  • it would be great to infer area/sub-topic expert based on self-rating of reviewers in their reviews to see where reviewers are in the cheriton scale (1..100, where cheriton is 420). this would help assing conflict papers, poorly reviewed papers, last minute papers to such reviewers. [Balachander Krishnamurthy]

  • comment threading

  • From the "Assign reviews manually by PC member" page, you can get to "Assign reviews manually by paper," but not vice versa. [Benjamin Pierce]

  • Again for tweaking the assignment, it would be really helpful to have a single global view of the whole assignment, showing all available information and allowing papers to be reassigned at will. [Benjamin Pierce]

  • How about cc'ing the requesting PC member on the invitation request email? [Benjamin Pierce] -- won't cc, but maybe resend mail

  • Check that review tokens and external reviewer requests mix
  • When I'm sending mail to people via the website, I want to make sure that it's going to be CC'd to me. It would be helpful if the mail template either offered me a field to control who gets messages or -- perhaps sufficient -- displayed an un-editable indication that this was going to happen automatically. [Benjamin Pierce]

  • Similarly, when an external reviewer goes to the website to decline (or confirm!) a review, it should be more obvious who will see the message. [Benjamin Pierce]

  • As a PC member, I might be interested in discussions even if I am not a reviewer who is emailed a comment by default. It would be nice if we could search on "commented after 7/23" or display and sort on things like the number of comments and the last time a comment was entered. [Fred Douglis]

  • It would be nice if the chair email generated when a review is updated included a diff. (Otherwise is it usually close to impossible to figure out what changed.) I would use this a lot, because I try to read all the reviews as they come in, to track how everything is going. [Benjamin Pierce]

  • Is there a way for me to sign up to see comments that are entered on every paper whatsoever? (Or does this already happen for the PC chair?) [Benjamin Pierce]

  • But maybe instead of sending a literal CC, the PC member could be sent a redacted copy [of the external review request] with the password blanked out? [Benjamin Pierce]

  • Mainly that the PC member can't see the wording of the request -- it feels like shooting blind. (More concretely, they can't see, for example, what deadline the subreviewers are being given.) Also, it means that they don't have a copy of the email in their files, so they can't do things like forward it with "PING" added to the subject line... :-) [Benjamin Pierce]

  • Here are some policy decisions that we changed by editing the code:

    Ability to prevent reviewers from seeing other reviews of papers they reviewed until shortly before PC meeting [Mark Gebhart, MICRO 41]

    Ability to show reviews to authors without showing comments to authors. We had some PC comments that were entered before the PC meeting that we didn't want to expose until after decisions had been made. [Mark Gebhart, MICRO 41]

    Not allow PC members to see anything about papers they didn't review until PC meeting [Mark Gebhart, MICRO 41]

    Allow PC members to view a paper's manuscript if they can view any information about a paper [Mark Gebhart, MICRO 41]

  • At the hotel where the PC meeting was held it was not possible to provide internet access to PC members. To allow PC members to view content during the meeting we used wget to spider the submission site a couple of days before the meeting as each member and burned a custom cd for each PC member. We had to make quite a few code changes to create a version of the site to spider that would not generate an large number of pages. We removed the ability to search and anything else that would depend on generating dynamic content. The resulting spidered webpages were around 100mb per PC member. This seemed to work very well for the meeting and we didn't have any trouble with people accessing their files on their machines. We found this to be quite useful and while it involved a fair amount of work the ability to turn off all of the dynamic content with a setting or flag in the code would be a great addition. We realize this may not be easy to do. [Mark Gebhart, MICRO 41]

  • It would be really useful to be able to turn off sending emails about comments to anyone outside of the PC. During the PC meeting, they naturally get used for all sorts of tasks like communicating final suggestions to the authors that reveal too much (e.g., the current acceptance state) to the outside reviewers. (I guess this can actually be achieved with settings already available -- right? So maybe this is a documentation issue...) [Benjamin Pierce]

  • it would be cute to have 1 config of hotcrp so that PC members get given as many scores drawn from a normal distribution as they have papers, and then can only use each score once. [Jon Crowcroft]

  • Do not log updates to reviews made using chair power, if those reviews are explicitly anonymous. [John Wilkes]

  1. It's somewhat of tedious to flip back and forth between per-paper and per-PC member review assignments. Could some better navigation shortcuts be made available? [John Wilkes]

Multi-round assignment process. Doing this showed up a couple of issues:

  1. Conflict checking

    • The list of names/words to ignore could usefully be both longer (e.g., add "Labs" and "research" to the list), and settable by the PC chair.

    • Conflicts of the form "username (affiliation)" leads to entire "affiliation" getting marked as a conflict, which seems overkill. Or was that the intent? The help text seems to suggest not. [John Wilkes]

Format checker: This worked really nicely. But ...

  1. It would be really helpful to be able to state "x% overage is OK" for the format checker, rather than editing the individual settings by hand, which tended to confuse people about what the real rules were. [John Wilkes]
  • it would be nice if there was an easy way to send email from one of a paper's pages to the author(s) or reviewers. As it is, it seems I have to go back to the home page, select Send Mail, select "Choose individual papers" and type in the paper number.

    Instead, how about a "send email about this paper" link on the "paper" page/tab which took me to the "send email" page with those values filled in? This is low priority in comparison to my previous request :-). [John Wilkes]

  1. The "manual by PC member" page gives no indication of discussion leads. [John Wilkes]
  • entry to identify presenting author

Minor features

  • Search: Assignments & conflicts: [External reviewer]
  • START-style enter-paper-all-at-once (no email verification)?
  • when email off: do not show authors to PC members, do not show reviewers to authors
  • scrub author information from PDFs