Recent site changes

by matt1 min read24th Jun 201180 comments


Site Meta

Recent site changes have generated more unhappiness than I expected. This post is a brief note to share resources that will make it easier for concerned site users to track what's happening and what we intend.

  1. First, know that we're listening. We'll make further site changes next week that will likely include some reversions.
  2. The official site issue tracker remains unchanged, but for the next week or so we'll work from this public Google Doc (just because it's lighter weight). Nothing on that document is a promise - just evidence of our current thinking. We'll strike out items on that list as we deliver them to our (private) staging server, and will roll them out onto the live site soon after.
  3. I've reached out to a small handful of SIAI and LessWrong heavyweights to track my balance as we make these changes. My feed should make it clear that I'm trying to act with calm rationality, but I'm obviously invested in the work we've shared to date and asking for some external help seems prudent.
  4. I'll track discussion on this post.


Some reflections:

  1. On what we did wrong
    1. We didn't predict the unhappiness our changes caused.
    2. We didn't make it clear that we were listening for feedback, and that changes were not final.
  2. On what many of you did right
    1. Calmly and politely voiced your concerns about some of our changes.
    2. Where you liked some of our changes, said so where we could see it. Thank you.
  3. On what many of you could have done differently, that would have increased average happiness (particularly mine)
    1. Calmly and politely voiced your concerns about some of our changes.
    2. Checked the very common assumption that the recent changes were final and would not be discussed. (Please note that I've first admitted fault in not making this clearer.)
A bare fact of this episode is that I'm feeling shellshocked, and have not enjoyed the experience of spending significant time and money trying to make LessWrong better. Unless my brain is built differently than others of my species this experience is likely to lead to spending less on projects like this in the future.
Please feel free to add to these lists in the comments, and please feel free to  other-optimize  me.  Tsuyoku Naritai.
… but try to be polite.


80 comments, sorted by Highlighting new comments since Today at 1:01 AM
New Comment
Some comments are truncated due to high volume. (⌘F to expand all)Change truncation settings

Recent site changes have generated more unhappiness than I expected.

I dunno. I'm pretty thrilled with the results. People usually only comment when they have a disagreement. When I write a big research post, almost every single comment is negative, and yet it gets 50 upvotes. So most people may appreciate the changes even if you only hear from critics.

Thanks for all your hard work, Matt! Try not to take the criticism too seriously. People get accustomed to something even when it's suboptimal, and then they complain when it's changed. But they'll get over it.

7tetsuo5510yI work in the customer support field and this is definately the case, you only ever hear the negatives.

Also, please actually pay attention to these requests, and don't add stuff that you don't know the community wants without talking about it first. In my experience, site redesigns can easily lead to large amounts of drama over very minor issues. If we're trying to be rationalist we should keep that in mind and proceed cautiously.

comment by JoshuaZ in the suggestion thread

I suspect much of the reaction is the feeling that many user suggestions and requests were ignored, while other actions were taken that users didn't necessarily ask for or desire.

I've gone through that thread to summarize all of the most popular suggestions. For now, I'm defining "most popular" as having > 10 karma so this list does not become even longer. (If anyone feels I've unfairly or incorrectly summarized a suggestion, please let me know.) Here they are in order, with their status as best as I can tell (Y for implemented, N for not implemented, ? for unsure).

Edit: The list takes up a lot of space when posted, so I'm hosting it here.

The totals are: 8 Y, 33 N, 7 ?. That's not a particularly stellar record. Now, personally I'm glad that some of these didn't get implemented (karma bounties and ... (read more)

9matt10yAgain, let me start by admitting fault in not communicating this more clearly: I assumed it was the most reasonable assumption that we would release features as they were ready, rather than releasing nothing, then everything. Once the redesign was ready we pushed it and all of the other features we'd finished to that point. There are other suggestions we've heard and plan to implement or are part way through implementing. … and, there are suggestions we don't plan to implement. See my notes on your notes at the end of our notes doc [] . Several requests contradicted each other ("avoid clutter" vs "add feature/button/info X"). For others… "make suggestions that will guide their redesign efforts" doesn't sound to me like a promise that we'll process the suggestion list in order. We considered all suggestions, and I tried to update my own opinion by comment upvotes. "Not a stellar record" seems a pretty harsh take compared to something like "considered all suggestions, and implemented their favourites". I hear you, and I think you may be accurately explaining why many felt annoyed/ignored… but "I made a suggestion and the suggestion wasn't executed" doesn't add up to good evidence that you were ignored. We should have made clearer in Louie's post [/lw/5by/official_less_wrong_redesign_call_for_suggestions] that we would use our own discretion in what we implemented, and that our budget was limited.
1Dreaded_Anomaly10yThank you for the detailed response. That modus operandi seems to be a trend in recent social web development (e.g. Facebook), but I don't think it's one that endears the developers to users. Prior communication is almost always a better option than just doing things whenever, especially in cases like this one where user advice and suggestions were explicitly requested. Also, I suspect the release of new features at the same time as new visuals contributed to users' inference that what's been done so far constituted "everything." If a new feature pops up by itself, it's just a new feature. The inclusion of the graphic redesign made it seem like this was the entire redesign, period. Thanks for the response. Before making a few specific replies, let me further explain why I went through the posted suggestions in that manner. Really, it was to make sure that I wasn't just upset because my own suggestions didn't appear to get much attention; I wanted to remove my own bias by looking at the contribution of the user base at large. As it turned out, my suggestions were repeated several times, and related suggestions were also repeated several times, as we both noted. On to specific notes: * I think you're interpreting #11 a touch broadly. When I see "clutter" in the context of web design, I think of something that takes up more space than its function is worth, i.e. a net negative contribution. It's not adding anything , it's just adding things that are not overall helpful/desired. Of course, that's my interpretation, and others may not share it. * I did not realize that the help page had been wikified, and that the welcome page will soon be as well. Thank you for pointing this out; it's a very good change. * I agree that #3, #14, #18, and #22 are complicated and expensive to implement. * I do not agree that #27, handling footnotes, is expensive. Collapsible sections take about 5 minutes to construct with CSS and Javascript, so the only d
6matt10yAgreed. As you and others have pointed out, a post like this one would have helped a lot. I have learned. For what it's worth, I did what you did, and we worked from a Freemind map very similar to your list, organised/weighted by upvotes, our estimate of effort required, and our personal desire to implement the features we worked on. I should have made that more public. I've been intentionally circumspect about the wikification, to try to leave the three wikified pages open as long as we can. I expect that soon enough the spammers will find them, and we'll have to lock them down as protected pages. For those paying close attention, see this [] . Code contributions are welcome [] :) It is my intention that agree/disagree has no karma attached. I think it very plausible that agree/disagree won't work as I hope it will for polling… but the test is as cheap as ordering which we implement first. Since we plan to implement agree/disagree anyway, we get the experiment for free. I agree that preview would be nice. What we have now works pretty well, so I'm judging this feature as low benefit/cost. Your suggested solution would be easy to implement, but I think would be confusing to users. Agreed. I have learned.
3Vladimir_Nesov10yDidn't notice this has gone live. Luke edited in the wiki source of about page [] what I think is just terribly wrong, and now it's the site's about page [] , but it's 5:40 AM and I can't figure out a way of undoing the problem without removing the content. From the page: These are not norms, these are skills, and there is no hope or point in enforcing them, particularly absent understanding the purpose. It's like saying that believing that ZFC is consistent is a community norm in a math seminar. Doesn't sound at all right.
0lukeprog10yI was not a lone actor. The new About page I wrote was approved by SI. The community norms are both norms and skills. I have seen them enforced by the community hundreds of times.
0Vladimir_Nesov10yI've just edited out the phrase "community norms" from the "about" page.
0lukeprog10yI like what you've come up with. Thanks.
-1Will_Newsome10y(Which to some extent does indeed mean you're not a lone actor, but it's worth noting that SI approval isn't much of an argument in favor of the About page's quality. (Just wanted to make sure people didn't assume you were making an argument from authority.))
0matt10yI assume "I can't figure out a way of undoing the problem" speaks to finding the right words for the about page. The technical way to solve the problem is to edit the wiki page (with the right words, whatever those may be), then wait a few hours or be logged in to LW and click the "Force reload from wiki" button at the bottom of the about page.
[-][anonymous]10y 15

A bare fact of this episode is that I'm feeling shellshocked, and have not enjoyed the experience of spending significant time and money trying to make LessWrong better. Unless my brain is built differently than others of my species this experience is likely to lead to spending less on projects like this in the future.

I find it funny that you almost need to apologize for having normal human emotions. Don't worry, I think no one holds this against you.

And if you need something positive to focus on, note that you're receiving positive karma for most of your interactions with the community.

Also let me say that the overall site design seems cleaner and better readable, I perhaps have one or two minor issues but overall you have indeed optimized LW. Since this is my favourite site online, let me give you a heartfelt thanks. :)

I know practically equivalent messages have been relayed by others, but I also know that our brains work in a way that makes a +1 less relevant than reading a new message with the same content.

I find it funny that you almost need to apologize for having normal human emotions.

Sounds very sensible to me. Status quo bias, complaining more than praising, being a jerk to whoever caused changes, and escalating drama stem from normal human emotions, too. Matt wins the moral high ground points and unlocks the special Postel move for... well, many things, which include questioning the "They're all ungrateful jerks!" reflex.

My apologies - my first and second comments on the changes were both uniformly negative, even though I find the changes to have been a net positive. (Now especially that the number of comments is again visible on the index pages, I no longer have any major gripes.) From having both personally done and seen friends do similar projects before, I should have known to provide much more positive feedback from the onset.

In particular, I like the general appearance of the new layout, the karma bubbles being expanded (for the first time in ages, I'm aware of my own karma changes again!), and the highlighting of new comments. The highlighting especially is a great change.

Anyone putting in this much work in a project like this would deserve to have a page full of praise and positive comments in addition to polite, constructive criticism. I'm very grateful for you doing all of this, and have immense respect both for your work on the actual update, and on your calm and reasonable manner of handling the unfairly harsh reaction.


I think you are doing a great job,in both implementing the changes and responding to the criticism. Though there have been some issues, that is normal for this sort of software upgrade, and I am pleased to note that my biggest concerns are reflected in the "immediately" section of your issues list. I'm sorry I haven't been vocal about my favorable opinions or my confidence that you can and will fix the problems. As a software developer myself, I know how discouraging it is when the users don't seem to appreciate your efforts.

Everyone who liked the update should perhaps vote up the post to cheer up our hard-working benefactors.

Woo! Comment #s are back. This pleases me, cheers for that :)

I think it would've helped if you had made a post about the changes when they were made, with a list of what had been changed and why and a clear statement that you were looking for feedback to modify or reverse changes that didn't work. That post would've been a place for people to give you kudos for all your hard work, and it could have encouraged a discussion that was more constructive and less antagonistic.

People naturally get put off when a site that they're comfortable with is suddenly different, and attention naturally gets drawn to the little annoyances. It can help to have a list of what's been changed to make it all more understandable, a promise that bugs and annoyances can be fixed, and examples of cool/helpful changes that help balance out the annoyances. A lot of that came out eventually in the discussion to the posts about the changes, but you missed your chance to set the tone right from the start and so the discussion had more of the feel of a user revolt rather than seeming like part of the process.


Thanks so much for putting all this work into the site. The fact that you're listening to advice, and even changing back things that you've put hard work into, will really go a long way to make LessWrong better. Thank you.

I find it amazing that none of all the extraordinary claims made on this site has ever triggered a similar amount of criticism than an almost completely irrelevant change to the layout.

Criticising claims needs evidence and arguments, criticising layout needs only preferences.

6Kaj_Sotala10yThat's a common pattern [] .
4timtyler10yThe SIAI and their views get their fair share of criticism here, IMHO.
1Normal_Anomaly10yI agree. Also, while reality is not graded on a curve, SIAI certainly gets more and better-received criticism than the vast majority of people/organizations get on their own blogs.
2wedrifid10yHypothesis: Measured in terms of either comment count or word count there has been more criticism of the unusual claims of the SIAI by XiXiDu alone than all discussion of the layout change. (Especially if limited to changes to the layout , as per XiXiDu's claim but also if we include all discussion of changes to user behavioural abilities, which seem to be far more controversial. ie. There is a significant chance that your premise is false.

Any thanks and complaints about the result of these changes aside, I want to unconditionally thank you for the hard work you've obviously put into making LW better. I'd like to help you feel more positive about your efforts: do you have an amazon wishlist? a tip jar? a favorite baked good I can send you? a novelty t-shirt you really want but feels just a bit too expensive?

8matt10yThanks for the thought - any tips to SIAI [], but let me know it happened so I get to share the warm fuzzies :)
0jsalvatier10yWould a Robin Hanson hour [] be a rewarding experience for you? I am the proud owner of one, but don't have anything that I really want to discuss.
1matt10yThanks and yes, but I qualify for Robin's offer myself, and claiming two hours might come across as rude :)
0jsalvatier10yhaha, I don't think it would be rude. Let me know if you change your mind.

Thanks for your hard work. I expect that once all the little snags get worked out this will be a material improvement over the old design.

Nobody is immune to various manifestations of status quo bias. Any change is likely to create controversies. The reaction is about as strong as I had expected before I knew what changes would be implemented. So, perhaps you made a mistake in underestimating the strength of criticism, but that is not much relevant to the quality of changes itself. Even if the changes were optimal (relative to the users' preferences), the reactions would not be much different.

That you are listening to the reactions is something which could be rationally assumed, and again, I... (read more)

[-][anonymous]10y 5

I wasn't sure of the right place to express it before, but thanks for all of the hard work you and the team have put into the redesign!

I trust the compliments more than the complaints, because there are more likely ways to negatively speak too soon than positive ones. The functional comments are probably of more value. I'll let you know in a week how I feel about the visual aspects, because I don't trust my judgment right when something familiar is changed.

People should also be aware of the relevant My Little Pony episode about design preferences, Suited For Success. Serious business.

1Normal_Anomaly10yThanks for the My Little Pony link. It's definitely relevant here, and I at least need to be reminded of that.

As I get some distance from seeing things for the first time, I definitely had some huge status quo bias and didn't properly account for how I'd learn the new icons - and there's also the fact that my old browser cache caused big bugs for the first few hours I saw the new site, and that was the period when I made my comments about it.

I bet this sort of thing is very common, and can think of two ways to handle it. One would be not caring, since people will be fine with it in a while anyhow. The other would be marketing and laying groundwork, not like a di... (read more)

My suggestions for revoking-the-comment feature:

  • Don't actually prevent voting, just limit the extent to which the comment can be downvoted. Say, no lower than -4. Or at least -2. Otherwise, allow voting. Maybe the community decides that it's actually a good comment! (Another variation: allow unlimited downvoting of a comment, but make it have no impact on Karma. It's probably an even better alternative.)
  • Don't indicate that a comment is revoked by striking through all its text. It makes it hard to read, which is annoying. It also makes it impossible to a
... (read more)

The comment counts are back on the list of topics. Huzzah!

Thanks for your hard work, and especially for the green borders on the comments added since the last time the page was viewed.

I think the mere fact of having unread comments highlighted in green more than makes up for the rest of the arguably worse changes (the old header is the only thing I regret (and frankly who cares, that's textbook bikeshedding), the other small changes are probably improvements).

As a fellow programmer who has to deal with usability and fickle users (though I make games and not websites), I can guess how you feel - it is annoying how people tend to freak out over small details that are obviously small and easy-to-fix bugs (like the cache issue), and weight ... (read more)

5Unnamed10yThat's my favorite too. I've gotten into the habit of following discussions by using the Recent Comments page, since it has been so hard to follow what's new with the threaded commenting system, and having new comments highlighted is going to make it much easier to follow the flow of the discussion in context. There's one issue that I noticed with that feature - viewing an individual comment gets counted as viewing the whole post. So when I click on a comment in Recent Comments, and then click "View the original post" to see the rest of the discussion, there aren't any green outlines so I don't have an easy way to tell which of the other comments are new. I wonder if it could be set up so that viewing an individual comment doesn't count as viewing the post.

So, I have just discovered how useful the "new post glow" feature can be. Thanks for implementing that!

[-][anonymous]10y 3

The nearby meetups feature is a great idea, but the current implementation creates a new challenge: All regular meetups are now making posts every time they have a meetup. This is more work for them, but more importantly, it's serious clutter on the discussion page.

Possible solutions: 1) Implement a calendar that can handle repeat events. 2) Add an option on the Add New Meetup tab which would allow users to either create a new post (what it does right now) or to link to another url (old post or wiki page). This would at least handle the clutter problem.

4matt10y1 is planned but unscheduled, and the discussion area post was a quick hack to give us comment discussion for meetups - our first implementation didn't support comments on meetups at all. … which is to say, I agree, and we plan to fix it. We figured that there are some advantages in the new system that made it worth publishing.

Can you change the roll out process? Maybe with a publicly accessible beta, and for bigger changes the chance to use the old version for a while after roll-out? My guess is that this is possible as long as the database is not changed.

In the two weeks since the site redesign, the front page has completely fallen off Google for keyword rationality. I would take a hard look at how the main page is being seen by search engines and optimize for a short list of keywords that matter. You can start by adding the phrase refining the art of human rationality to the title.

Currently, the only thing for me that is close to a dealbreaker is the issue about not being able to delete comments. If that ability is restored, I should be mostly complaint-free for the time being.

One feature that I would like to request, however, is the ability to more easily retrieve old private messages -- perhaps separate inboxes for comments and PMs.

1matt10yWe're working on restoring delete for childless comments now and should roll it out this week. That will probably be a two step process - retract → delete. I understand that User:Alicorn is working on making the case that we should allow delete for comments with children. We have no current plans to work on improving PMs. We agree it'd be cool, we're just not excited about it (but… it's open source [] - don't wait for us).
2Alicorn10yIf I can possibly convince you to do anything PM-related, please have a separate PM inbox. It is nearly impossible to find old received PMs. Usually I have to sort through sent PMs, find the date, and then click-click-click-click my way through my inbox history until I find the right time. This is frustrating.
0timtyler10yEr, that might help - thanks! Update 2011-07-04: You fixed it - yay! You now have to refresh for it to work - but no big deal.
0[anonymous]10yYou fixed it - yay!

Ooooh, looks like everything I disliked about the update has been fixed (last one sometime in the lest few hours it looks like), so consider any complaints I had withdrawn. This In now officially awesome and you're awesome for making it! ^_^

Thanks for putting in a lot of work and helping make the community better etc. etc.

Edit: Wait, what, now it's back to the bad colour scheme again?!

0Armok_GoB10yWait, what, now it's back to the bad colour scheme again?!
0Armok_GoB10yWait, what, now it's back to the bad colour scheme again?!
0Armok_GoB10yWait, what, now it's back to the bad colour scheme again?!

I feel that the main problem, perhaps the ONLY problem, is not among the ones you listed: The changes were made involuntarily, without any warning. What you should have done instead was somehting like making the changes, but have them as an alternate mode so that you could chose if you wanted to use the new or old design at first. Then, once you had sorted out almost of the issues you could do things like having the new design be the default for new users, changing everyone profile to it but allowing the option to change back, etc. Only once almost everyon... (read more)

9Emile10yAw c'mon, there was plenty of warnings, the various thread on improvement suggestions, the screenshots ... I don't think I would have preferred to have even more threads devoted to css bikeshedding.
-2Armok_GoB10yIf there were warnings, I didn't see them, and I miss few things around here.
6Kaj_Sotala10yThere [] were [] several [] warnings [] . Personally I'd gotten rather impatient waiting for the actual change.
5Armok_GoB10yI've seen those. They were all posted AGES ago, with no indication of whenthe changes would ahpen, and were also based of erly planing stage mockups and not the actual project. A proper warning would have been "Hey, we've finished the redising and are rolling it out tomorow unless there are lots of objections. it'll look like this: actual screenshot "
7handoflixue10yFor what it's worth, I almost never venture in to the discussion area, and my reaction was immediately "Ahh, the new changes must have gone live!" That said, for anyone who doesn't touch Discussion, or objects to the "live testing" philosophy, I can see it being an issue. On the other hand, setting up an entire cloned test site, or doing all the changes via CSS, would have made this a more complex project - especially given that they're not being paid to do this, I consider it pretty reasonable to do "live QA". It also has the added advantage of getting feedback from everyone, instead of just the Favored Few of the Bayesian Conspiracy :)
6Armok_GoB10yThe difference between "Ahh, the new changes must have gone live!" and "Ahh, some new changes must have gone live!" is negligible in this case I think. The problem was that you didn't think before you opened up the site "The new changes are probably up by now" and thus there's still a surprise involved. And thanks for finding a name to where our disagreements lie. I object strongly to live testing.
3handoflixue10yYou are not alone there. I think that a failure to communicate about that was the biggest issue here, and doing it at all was the second.
2Kaj_Sotala10yFor what it's worth, I too was surprised, but it was a pleasant surprise. "Ah, the changes have finally gone live, awesome!"
4Emile10yThe last warning was less than a month before the changes (and in my memory is "pretty recent"), and had screenshots pretty close to what we have now, though some suggestions given in the comments were taken into account (use thumbs for vote up/down not for agree/disagree). I for one thing matt (and others?) were right to deploy the new design as soon as possible rather than spend more time bikeshedding. Yes, maybe they could have deployed a beta somewhere public, but that would have taken time that would have been better spent improving the site.
4Armok_GoB10yLooks like we simply disagree. A month IS a very long time according to me. And I think not pissing of the users would have been worth the extra time spent.

I'm really liking the meetup system. Good job.

Perhaps having a public test server would make it easier to get continuous feedback?

First off, let me say thank you for all the work that's gone into the site update by everyone involved! The three changes I like most are the new header design (especially the clear separation between Main and Discussion - the old menu was too cluttered), the nearby meetup section, and the expanding karma bubbles.

I had one question about how the nearby meetup list and Location. Is the meetup list supposed to sort by location somehow? If so, what do I need to put in my location? Thanks!

4matt10yIt should be locating you by your IP address and showing the 5 nearest meetups from those scheduled over the next 14 days. We think it's broken. We intend (implementation unscheduled) to use the location you enter in your preferences if you have supplied one.
0Yurifury10yCould you use location services to determine their location? For people who have not explicitly set a location, a button is displayed. When it is clicked, it can ask the browser for location data, and determine their suburb/country by using a service like this [].
0Emile10yIf the implementation is simpler, maybe a tag-based watchlist would be better - you could pick a list of tags to watch, meetups would be tagged by continent/county/city (whatever makes the most sense), and you could edit the list of tags you watch. OK, maybe that would be more complicated. Just brainstorming :P

Something I would really like is for the main site and the "main" button at the top to go to the new posts page, not just the promoted posts. Something only gets promoted every few weeks, and I want to see all the new content, not just the best. I realize that I may be in the minority on this, so I won't mind if you don't implement it because you think it would be ill received.

Also, can the green glow around new comments be extended to posts that contain new comments? I can't always remember how many comments a post had at last glance, and I'd like to see where there are new ones.

0matt10yon promoted posts: You'd have to convince Eliezer on that before we'd act. I'd vote against it - I like the curation he provides. on posts with new comments: We have something planned there, and while not actually locked into our schedule, I'd love to get it out Real Soon Now™.

Thanks for your hard work! I'm really happy with the changes for the most part, especially the green borders on new comments and your quick response to the request I had in the last thread about long usernames. As someone who's just barely a programming novice, I understand how hard making some of these changes would have been.

As for the stuff in the Google doc and this thread, it all looks like good ideas and I'm looking forward to seeing it implemented. My only real problems are missing the old header and missing the "comments" links that used to come with every post.

Here's my number one comment, and the one thing I'd want to see modified:

  • With the current style, the "Upcoming Meetups" change has squeezed down "Recent Comments" and "Recent Posts" to the second page. This is disastrous for my participation in the site -- I expect a drastic reduction in my participation for the mere reason that I can no longer see at a glance if there's a new comment or a new post, that I might want to respond to.

Secondary comments:

  • Keep "Vote up" "Vote Down" as icons, but move them n

... (read more)
0Jonathan_Graehl10yThe icons are all at the bottom (good) and have tooltips (good). The reply icon is conventional, so I don't mind it. However, it's smaller (narrower) than it could be (whether it has text or not). I understand the voting use case you have in mind; some people may read the comment, then check its rating and vote only if it's "too high" or "too low". And there's a certain logic to having the up and down near the score it will affect. If you vote independently of the current score, it's fine.

I think I just broke the meetup submission function. Sorry.

Feature request: it would be great if retracted comments were initially rendered "folded" like low-scored comments, and had to be explicitly unfolded.

2Oscar_Cunningham10yI believe folding also folds their children, which could be irritating.
0timtyler10yA "collapse-to-title" option would sometimes be useful for low-scored comments, previously-read comments and retracted comments. It would need to not collapse the children, obviously.