I support this request.
Quantopian forum should have all public algo searchable by keywords at least in 3 category and as fast and accurate as it's grandfather Whealth-Lab Developer done:

Title contains _____________
Public algo code contains _____________
Author is _____________

Regarding sharing algos, my 2 cents is that the algo should first be put under revision control in something like github. Then, instead of posting the algo to the forum for copying, a link to the repository would be provided (with a snapshot of the backtest/notebook results). By automatically setting up posted algos in a repository, they would be ready up for collaborative improvement, organization, forking, etc.

Also, at QuantCon, in a conversation with one of the participants, it was pointed out that Quantopian is not set up for teams to collaborate effectively (he pointed to http://www.kaggle.com/ as an example). In theory, one could set up something on github (e.g. https://github.com/gkiehne/quantopian), but without integration with the Quantopian backtester/research, there is a lot of copying/pasting, etc. A collaborative coding and computing environment would help move beyond the concept of individual hedge fund "managers" and their code developed in isolation, to team-based efforts.

To be honest I would rather have the developers focus on the platform :) but the forum and the community is important also. I would probably voucher for them to get a few people from community and install a vbulletin forum in a different host and hand them the Admin / Moderation.
And move algorithm sharing to an external source. Important is to be able to publish a snap-shot of the backtest maybe auto generate a report that can easily be posted on forums.

Wealth-Lab! Only reason I learned Pascal was Wealth-Lab. Them getting bought out by Fidelity was a sad day for young me.

@ Lucas da Silva: exactly what I was thinking of !

As always, lots of great ideas here.

Adding new features to the forum is getting close to the top of the priority list for the team. I'd unify a lot of the feature requests above under a single umbrella: quickly surface interesting and relevant content for each visitor to Quantopian. I love the idea of tagging for that. I also think that we will benefit from labeling "new member" questions in specific. I'm not sold on actual forum separations - I think we get a lot of serendipity by having everything in one forum.

We'll be making progress on this, I expect, in late spring and summer.

My 2c, couple of thoughts:
Categorizing/tagging from users would be difficult to write, and on the other hand, categorizing by Q would be time-consuming and likely receive complaints.
Simplest step forward would be an incremental change, simply a thumbs-up icon on each post and/or page and some means of listing by those up votes.
Also, clicking on someone's name/handle, simply expand the lists of their posts and algos, bingo, that's done.

Here's a more advanced concept along the lines of finding info: Maybe some of you have noticed this particular problem on youtube, you can sort (filter) by view count, however that's often not really all that meaningful, there can be a video going viral with very few views because it is newish so it could be at the bottom of that view count list. What would make sense is views per day. Mmm? Let that sink in. Suddenly you have a metric that is useful. So maybe thumbs-up per month. :P Views per day.

Making large amounts of information as useful as possible, great goal. A small yet important part of that is:
Folks ought to adopt into their world a simple search technique via external search engines: Use site:quantopian.com
That limits results just to the quantopian website. For example, to the original post, there are 15 results that one can breeze thru fairly quickly for "mean variance" like so: https://www.google.com/#q=%22Mean+Variance%22+site:quantopian.com

Add the word optimizer and then you have only 7 results. In-house search, tough, google easy, you could just switch the search here over to use of google (code they supply) and search could be considered done.

Again, to address part of the original post, 7 results for "yahoo financial": https://www.google.com/#q=%22yahoo+financial%22+site:quantopian.com

Understand that the content of attached algorithms (the code) sits in the html of the page so search engines pick that up. These results ought to all have examples of actual fetcher code, small example: https://www.google.com/#q=%22fetch_csv%22+site:quantopian.com

wish list :)
1. voting up Grant's github configuration management + an easy way to submit an algorithm from github for backtest.
2. voting up Gary's google search instead of the current search
3. "stack exchange" style of Q&A/voting/badges for the blog
4. An easy way to control Quantopian backtest inputs to an algorithm for test vector generation
5. Ability to copy out the backtest log for analysis (maybe the issue is chrome)

I disagree with the idea of a traditional web forum platform, the UX with those is generally very bad since their interfaces are usually cluttered with tons of useless info. On the top of my head i recall different horizontal menus, count of thread, count of posts, list of admins, list of users, list of active users, "motd" per forum and sub-forum, search forms (sometime replicated at different levels), and generally a clumsier presentation of contents. Oh, and -God forbids- users' signatures.

Although this Community section of Quantopian site is different from what an user is used to I have to say that it's very neat and clean, it's quite good to read, also it's splendidly integrated into the rest of the website and I guess the bonus of being able to attach backtest to a thread is partially due to that.

If anything I'd suggest to make the current Community a little better with the following changes:

1. implement advanced search operators
   
2. implement ability to tag threads
   
3. make the width limit higher to adapt to bigger windows
   
4. implement anchors in threads

1) refers to advance operators à-la-Google, to name a few I'd implment
title: search only in title (no operator would search in body and title)
author: search the author's full name
is:[post|backtest] to search different type of contents
has:backtest to filter posts with a backtest attached
before:|after: to filter dates (please RFC 3339 here, i.e. YYYY-MM-DD, I already talked with someone on Feedback about this)
and|or|not logical operators
"" literal search, respecting spaces and extra characters

These tools alone provide an extreme flexibility while keeping the UI very clean ("advanced search" page of web forum anyone?), also those should integrate pretty esaily with the current site model.
Btw is the Quantopian website based on an open source or going open source by its own in future?

2) in past few years I've seen more and more social sites moving from fixed forum/sub-forums/sub-sub-forum taxonomy to a tag|label approach, I suggest to attach another little piece of information to discussion (or possibly single posts) to easy categorize them. This is a better approach because sometimes a post overlaps different topics, with a fixed scheme you have to sacrifice one, with tags you tag it twice. There's no real separation, you extract the posts you need from a single cauldron.
Along with this goes
tag: filter only specified tag

3) this is a lesser suggestion, make the message colum of posts grow over current 1014px limit
4) another lesser suggestion, implement HTML anchors in replies where the time of the submission is, this would allow to link to a specific reply and have it presented on top of the browser page when the link opens.

edit: accidental submit before completion

edit: had a little hard time with the numbered lists, what syntax is this?

I second the idea to incorporate version control. As it is I save all my work on the side in a Git repo so I don't lose any of my work.