Link-it: The nextgen reddit written in Common Lisp is online

17 views
Skip to first unread message

Marc Battyani

unread,
Dec 7, 2005, 4:41:26 PM12/7/05
to
The second release of linkit is online at the usual place:
http://linkit.fractalconcept.com
Be sure to register to see the new functionalities like the positive and
negative filters for categories and submitters and the ajax like stuff.

It is what I have so far after about 7h of work.

I don't know what do you think, but for me that the proof of concept is over
and conclusive.
(I hope that the other people who are doing one will be able to show
something soon.)

Now what next ?
We can stop threre or we can go on to much interesting stuff based on that
platform.

If we go on, obviously the ranking of links and submitters is the next step.
I did not implement it yet as maybe we would like to use a different system
of ranking.
For instance we could try to make correlations between actions of users and
present "selected for you by linkit" links on a personalized basis.
We could register votes and let users assign more or less weight to voters.
Or discovers the weights magically.

As for new functionalities, I've already added positive and negative filters
for categories and submitters as it seems ovbious to me that not anybody is
interested in US politics for instance.

What about user comments about links ? Yes, no, maybe ? Yes with the same
show/kill filters

A new interface ? Other ideas ?

Marc


Peter Seibel

unread,
Dec 7, 2005, 5:06:47 PM12/7/05
to
"Marc Battyani" <Marc.B...@fractalconcept.com> writes:

> The second release of linkit is online at the usual place:
> http://linkit.fractalconcept.com
> Be sure to register to see the new functionalities like the positive and
> negative filters for categories and submitters and the ajax like stuff.

I must be missing something. After I click on the Submit link I get a
form but there's no submit button and hitting Enter after I type the
URL doesn't seem to do anything. How do I actually submit a URL?
(Tried in both Firefox and Safari on OS X).

-Peter

--
Peter Seibel * pe...@gigamonkeys.com
Gigamonkeys Consulting * http://www.gigamonkeys.com/
Practical Common Lisp * http://www.gigamonkeys.com/book/

Wade Humeniuk

unread,
Dec 7, 2005, 5:14:25 PM12/7/05
to
Peter Seibel wrote:
> "Marc Battyani" <Marc.B...@fractalconcept.com> writes:
>
>
>>The second release of linkit is online at the usual place:
>>http://linkit.fractalconcept.com
>>Be sure to register to see the new functionalities like the positive and
>>negative filters for categories and submitters and the ajax like stuff.
>
>
> I must be missing something. After I click on the Submit link I get a
> form but there's no submit button and hitting Enter after I type the
> URL doesn't seem to do anything. How do I actually submit a URL?
> (Tried in both Firefox and Safari on OS X).
>
> -Peter
>

It used to work, I am getting the same problem now.

Wade

Creighton Hogg

unread,
Dec 7, 2005, 5:02:35 PM12/7/05
to

Hi Marc, can you explain the submission interface a little
bit? At the moment, it's not very clear what's going on.
Are you in the middle of changing it, because I can't even
try to submit anything?

Marc Battyani

unread,
Dec 7, 2005, 5:20:04 PM12/7/05
to

"Wade Humeniuk" <whumeniu+...@telus.net>

Sorry, I've just changed the user rights. Only the admin can submit now. :(
I correct this.

Marc


Marc Battyani

unread,
Dec 7, 2005, 5:23:22 PM12/7/05
to

"Marc Battyani" <Marc.B...@fractalconcept.com> wrote

Corrected.
People doing web development without a listener are missing something great!

Marc


Peter Seibel

unread,
Dec 7, 2005, 5:35:47 PM12/7/05
to
"Marc Battyani" <Marc.B...@fractalconcept.com> writes:

So, does clicking on the arrows actually do anything or is that not
implemented yet? Everything seems to have zero points.

Marc Battyani

unread,
Dec 7, 2005, 5:44:12 PM12/7/05
to

"Peter Seibel" <pe...@gigamonkeys.com> wrote

>
> So, does clicking on the arrows actually do anything or is that not
> implemented yet? Everything seems to have zero points.

Yes, I have not implemented the rating system yet. (see my first post)
Before doing the same rating as reddit, I wanted to know if people had
better ideas.

Marc


ajones

unread,
Dec 7, 2005, 6:38:28 PM12/7/05
to

My take on a rating system (kind of long):

* Give a bonus to stuff that is new. I think simply having a "new"
category is the best way to do this, but consider it for other sections
as unread stories are more useful than read ones.

* Put in an option to omit visited links. If I read it somewhere else
on the internet I probably won't want to see it on linkit. If I've
already opened it once from linkit then why would I do it again? The
only reason I can think of is for repeat reads, which are pretty
unlikely, but can always be handled with a "stuff I have read" section.

* Allow category rating modifiers. If I want to read articles about
tomato-based painting, I should be able to give those articles a slight
bump in spite of the downmods they may get from the less
tomato-painting enthused. Obviously this works two ways, and should
have both relative and absolute modifiers.

* Put a more than linear effect of submissions from users you rate
consistently. This should probably be pretty small, but the more
someone does a good job the more the system should reward their work. I
doubt exponential change is the solution here, but something where
getting consistent moderation earns you more points is probably
worthwhile.

* Provide recommendations from other people's likes by category only.
Evidently a lot of people on reddit that like some of the same articles
I do (recommended links) are also interested in politics and religion
of some form or another. I come to reddit for articles about
programming and startups, not politics and religion.

That said, here are some sample computations:
New articles:
Descending by date/time.

"Hot" articles:
Rating + Category modifier(s) + (overall moderation of user - 30)^5/4 -
(Time about 30 days ago - time of posting)

(overall user moderation if above/below 30 points)^5/4
30 and 5/4 are arbitrary numbers here, but generally are intended to
ensure a consistent rating pattern and limit any drastic effects of
this score modification respectively.

Subtracting the time 30 days ago from the time of posting, then
removing that from the score adds points for new stories (more likely
to be hot) and removes increasing amounts of points for older stories.

My big concern with this detailed of a rating system is that the
influences involved either affect the score too much or too little, and
that this will turn into a matter of individual taste.

Of course, you could always build in a facility for people to create
their own ranking algorithm and just supply a sensible default.

ajones

unread,
Dec 7, 2005, 6:51:45 PM12/7/05
to
If you implement nothing else though, there is one feature that I think
is absolutely necessary:

Find a way to prevent double posting.

Given a large enough database it would be impossible to check every
single link submission at the submission time, but there are ways
around that. Maybe have a "double posts are stupid" moderation option
that flags the post for checkup. Having columns for domain and file
(everything up to the last slash) could track down a lot of it too.
However you do it this feature alone would make linkit worth more to me
than reddit.

hyr...@gmail.com

unread,
Dec 7, 2005, 7:25:23 PM12/7/05
to
> Given a large enough database it would be impossible to check every
> single link submission at the submission time, but there are ways
> around that.

Wouldn't some kind of hashing on URLs make dupe reduction/elimination
possible?

Creighton Hogg

unread,
Dec 7, 2005, 8:12:21 PM12/7/05
to
On Wed, 7 Dec 2005, ajones wrote:

>
> Marc Battyani wrote:
> > "Peter Seibel" <pe...@gigamonkeys.com> wrote
> > >
> > > So, does clicking on the arrows actually do anything or is that not
> > > implemented yet? Everything seems to have zero points.
> >
> > Yes, I have not implemented the rating system yet. (see my first post)
> > Before doing the same rating as reddit, I wanted to know if people had
> > better ideas.

<big snip>

> Of course, you could always build in a facility for people to create
> their own ranking algorithm and just supply a sensible default.

*This* is a pretty cool idea, I think. Also, would it not
be a rather cool demonstration of the extensibility of lisp
and functional techniques?

Emre Sevinc

unread,
Dec 7, 2005, 10:22:29 PM12/7/05
to
>>>>> "ajones" == ajones <ajo...@gmail.com> writes:

ajones> Marc Battyani wrote:
>> "Peter Seibel" <pe...@gigamonkeys.com> wrote
>> >
>> > So, does clicking on the arrows actually do anything or is
>> that not > implemented yet? Everything seems to have zero
>> points.
>>
>> Yes, I have not implemented the rating system yet. (see my
>> first post) Before doing the same rating as reddit, I wanted to
>> know if people had better ideas.

ajones> My take on a rating system (kind of long):

ajones> Of course, you could always build in a facility for people
ajones> to create their own ranking algorithm and just supply a
ajones> sensible default.

Wow! That, I would call a feature! ;-)

Providing users a few ranking schemes (with fancy names?) from
a drop down list, and also giving them the opportunity
to create a brand new ranking scheme, give it a name (and
an option to make it public) and save it. This sounds very cool
and innovative.


Happy hacking,

--
Emre Sevinc

eMBA Software Developer Actively engaged in:
http:www.bilgi.edu.tr http://ileriseviye.org
http://www.bilgi.edu.tr http://fazlamesai.net
Cognitive Science Student http://cazci.com
http://www.cogsci.boun.edu.tr

Sylvain

unread,
Dec 7, 2005, 10:27:37 PM12/7/05
to
Emre Sevinc wrote:

> Wow! That, I would call a feature! ;-)

alternatively, define an API so that an user can
write a ranking algorithm in lisp.

non lisper or lazy users could choose from either
a few pre-made algorithms, or that of other users...

--Sylvain

nall...@gmail.com

unread,
Dec 7, 2005, 10:51:15 PM12/7/05
to
>Providing users a few ranking schemes (with fancy names?) from
>a drop down list, and also giving them the opportunity
>to create a brand new ranking scheme, give it a name (and
>an option to make it public) and save it. This sounds very cool
>and innovative.

There was a post by ajones in the old thread that didn't get much
attention

>I've got one better, instead of just making independent websites +
>databases agree on an api now for the sites to share data with each
>other. Then if someone is too lazy to (submit to/register on/even read)
>multiple sites they don't have to.

Agreeing a flexible protocol for exchanging data and metadata would
enable anyone to experiment with new algorithms on data collected from
preexisting ones, as well as enable newer, more innovative algorithms
to be adapted quickly and guarantee they will be backwards compatible
with older data.

BTW, might I suggest having _three_ rating buttons: on for up, one for
down, and one for spam. That way metadata can be taken both from
specific likes and dislikes without spam screwing it up, also clicking
on the 'spam' button would cause that post to disappear from your view
of the linkkit.

nick

Creighton Hogg

unread,
Dec 8, 2005, 12:55:31 AM12/8/05
to

Well, if you wanted to take it to the next level of
complexity writing algorithms wouldn't even necessarily
require knowledge of lisp to make simple ones. You could
have a layer of macros to parse somekindof embedded language
that would look alot less lispy. I don't know if that'd
really be a good idea, but I think reaching as many users as
possible is cool.

cjs...@gmail.com

unread,
Dec 8, 2005, 3:11:18 AM12/8/05
to
hmm.. when i click any link on the opening page, i just get a blank page

Marc Battyani

unread,
Dec 8, 2005, 3:44:57 AM12/8/05
to

<cjs...@gmail.com> wrote

> hmm.. when i click any link on the opening page, i just get a blank page

Should work now. I tested an "improvement"... which was not ;-)

Marc


Message has been deleted

Marc Battyani

unread,
Dec 8, 2005, 4:54:40 AM12/8/05
to

"Marc Battyani" <Marc.B...@fractalconcept.com> wrote

I have seen that ther are already some posts with very interesting ideas and
propositions. I can't really reply to them now as I somehow have to do real
work. ;-)
But you can continue to propose ideas anyway. Also you can post code
snipsets to support your ideas. In Common Lisp obviously, or in javascript
for the interface.

I will reply to them this evening. For now I will just check it from time to
time to see if all is ok.

Marc


Tayssir John Gabbour

unread,
Dec 8, 2005, 8:33:22 AM12/8/05
to
Marc Battyani wrote:
> > A new interface ? Other ideas ?
>
> I have seen that ther are already some posts with very interesting ideas and
> propositions. I can't really reply to them now as I somehow have to do real
> work. ;-)
> But you can continue to propose ideas anyway. Also you can post code
> snipsets to support your ideas. In Common Lisp obviously, or in javascript
> for the interface.
>
> I will reply to them this evening. For now I will just check it from time to
> time to see if all is ok.

Some things I've noticed as a user; ignore any which are mistaken:
* Adding categories is very slow... (Maybe you're intending this to be
some wizzy javascript thing later, but right now giving them its own
page?)
* Nice just to submit urls without going to a special page.
* I'd like to stay logged in between browser visits.

Tayssir

Paul Wallich

unread,
Dec 8, 2005, 10:00:51 AM12/8/05
to
Emre Sevinc wrote:
>>>>>>"ajones" == ajones <ajo...@gmail.com> writes:
>
>
> ajones> Marc Battyani wrote:
> >> "Peter Seibel" <pe...@gigamonkeys.com> wrote
> >> >
> >> > So, does clicking on the arrows actually do anything or is
> >> that not > implemented yet? Everything seems to have zero
> >> points.
> >>
> >> Yes, I have not implemented the rating system yet. (see my
> >> first post) Before doing the same rating as reddit, I wanted to
> >> know if people had better ideas.
>
> ajones> My take on a rating system (kind of long):
>
> ajones> Of course, you could always build in a facility for people
> ajones> to create their own ranking algorithm and just supply a
> ajones> sensible default.
>
> Wow! That, I would call a feature! ;-)
>
> Providing users a few ranking schemes (with fancy names?) from
> a drop down list, and also giving them the opportunity
> to create a brand new ranking scheme, give it a name (and
> an option to make it public) and save it. This sounds very cool
> and innovative.

That would be particularly sweet if you let users switch between schemes
easily (e.g a bunch of url suffixes so that they could see different
slices of the site depending on their mood or interest of the moment.
(Sometimes I want to see a focused set of stuff, sometimes not.) Then
you could go meta and watch how many people used each ranking scheme,
and offer them to other people based on that...

paul

ajones

unread,
Dec 8, 2005, 11:42:29 AM12/8/05
to

Matthias wrote:
> "Marc Battyani" <Marc.B...@fractalconcept.com> writes:
>
> > The second release of linkit is online at the usual place:
> > http://linkit.fractalconcept.com
> > Be sure to register to see the new functionalities like the positive and
> > negative filters for categories and submitters and the ajax like stuff.
> >
> > It is what I have so far after about 7h of work.
> >
> > I don't know what do you think, but for me that the proof of concept is over
> > and conclusive.
> > (I hope that the other people who are doing one will be able to show
> > something soon.)
> >
> > Now what next ?
>
> Depends on what point you want to make. If you want to make it cooler
> than the competition the next natural step is to improve usability.
> These are seemingly simple things such as not reloading the complete
> page when a user rates a link. Or removing the http:// entry in the
> submit form. And encouraging the user to give _short_ descriptions
> for their links. (Maybe "title" should be "one line description" or
> so, the "title" can be inferred from the web page if needed at all.)

All good points. Another point of usability that I would like to see is
placing the title of the page in a title tag on the linik (so hovering
over it gives you the true page title).

>
> It's very impressive how fast you got the infrastructure working. But
> I don't think that the infrastructure is the hard part here. The hard
> part is the social engineering. (And that's not just the ranking
> function.)

I can see your point there. Sophisticated algorithms and a bazillion
features do not a good website make. Fluid design and a good community
are probably a lot more important, but considering who this site is
trying to appeal to (lisp coders?) having sophisticated algorithms and
a bazillion (optional) features is probably the best way to go about it
... after the social stuff is done right.

>
> Concerning comments and all sorts of cool features: I would introduce
> them very carefully, only when I'm almost positive they are a good
> idea. I'm quite sure that you are capable of building more features
> faster than most people. Unfortunately, this alone won't create a
> cool site.

Already said what I was about to. I got a bit carried away tossing out
features, but it was under the assumption that at least half would get
a "well thats just stupid" and another quarter would be marked as too
much to be worth it. I think a carefully chosen initial feature set is
the only way to do this right.

-Adam

pkh...@gmail.com

unread,
Dec 8, 2005, 3:31:54 PM12/8/05
to
Marc Battyani wrote:
> I have seen that ther are already some posts with very interesting ideas and
> propositions. I can't really reply to them now as I somehow have to do real
> work. ;-)
> But you can continue to propose ideas anyway. Also you can post code
> snipsets to support your ideas. In Common Lisp obviously, or in javascript
> for the interface.

First, wow. I can't express the amazement at the amount of work you're
managed to do in only a couple hours.

Second, my suggestion: Ideally, we'd want to use scoring information by
each user to tailor their link list, and show them links they are
likely to appreciate (probably because a group of similar users liked
it too). In other words, give greater credence to the scoring of user
who score other articles similarly as us. However, with only an
unidimensional scoring system, extracting meaningful information is
obviously harder than with a more expressive scoring system. So, how
about scoring with a tag and an up/down arrow? Something that allows me
to express "tag: Good-Lisp-Language; score: down", meaning that it's
about (good) Lisp and I don't like it [implying that it's because it is
an element of Good-Lisp-Language]. Note that the tag system allows to
decouple preferences and judgement of quality. It might be better to
actually have the ability to score both the article and the relevance
of the tag separately. With this system, one could only vote a tag up,
not down.

Finally, the eternal problem with tags is that we end up with tons of
irregularly-named, unclear, redundant or overlapping tags. The clarity
and regularity of the names should probably only be enforced by having
a style guide for tags, and not using tags of bad style. As for
redundance and overlapping, some dynamism and human laziness could
help. When scoring an article, a popup (yet more JS :/) could appear,
allowing one to choose popular (in general, or for that entry) tags, or
to enter one in a form (here, something like Google Suggests could help
alleviate further the problem of bad tagging).

Paul Khuong

Marc Battyani

unread,
Dec 8, 2005, 6:37:10 PM12/8/05
to

"ajones" <ajo...@gmail.com> wrote

> Marc Battyani wrote:
> > "Peter Seibel" <pe...@gigamonkeys.com> wrote
> > >
> > > So, does clicking on the arrows actually do anything or is that not
> > > implemented yet? Everything seems to have zero points.
> >
> > Yes, I have not implemented the rating system yet. (see my first post)
> > Before doing the same rating as reddit, I wanted to know if people had
> > better ideas.
>
> My take on a rating system (kind of long):
>
> * Give a bonus to stuff that is new. I think simply having a "new"
> category is the best way to do this, but consider it for other sections
> as unread stories are more useful than read ones.

There is already a "new" list.
Also the "hot" list will be filtered with a filter that takes the
"freshness" into account

> * Put in an option to omit visited links. If I read it somewhere else
> on the internet I probably won't want to see it on linkit. If I've
> already opened it once from linkit then why would I do it again? The
> only reason I can think of is for repeat reads, which are pretty
> unlikely, but can always be handled with a "stuff I have read" section.

Yes it's already there but not yet operational. I have to put some
javascript to notify which links are clicked

> * Allow category rating modifiers. If I want to read articles about
> tomato-based painting, I should be able to give those articles a slight
> bump in spite of the downmods they may get from the less
> tomato-painting enthused. Obviously this works two ways, and should
> have both relative and absolute modifiers.

I've thought about this one and I think the bonus/malus should be given in
percentage.
ie lisp:+ 20%, politics:-20%, Python: hardkill

> * Put a more than linear effect of submissions from users you rate
> consistently. This should probably be pretty small, but the more
> someone does a good job the more the system should reward their work. I
> doubt exponential change is the solution here, but something where
> getting consistent moderation earns you more points is probably
> worthwhile.

I wanted to add a bonus % like for the categories

> * Provide recommendations from other people's likes by category only.
> Evidently a lot of people on reddit that like some of the same articles
> I do (recommended links) are also interested in politics and religion
> of some form or another. I come to reddit for articles about
> programming and startups, not politics and religion.

There is already a categories filter. isn't it enough ?

> That said, here are some sample computations:
> New articles:
> Descending by date/time.
>
> "Hot" articles:
> Rating + Category modifier(s) + (overall moderation of user - 30)^5/4 -
> (Time about 30 days ago - time of posting)
>
> (overall user moderation if above/below 30 points)^5/4
> 30 and 5/4 are arbitrary numbers here, but generally are intended to
> ensure a consistent rating pattern and limit any drastic effects of
> this score modification respectively.

Well why not, those numbers must be tweaked with real data when it is
available.

> Subtracting the time 30 days ago from the time of posting, then
> removing that from the score adds points for new stories (more likely
> to be hot) and removes increasing amounts of points for older stories.

Users should be able to set a too old limit

> My big concern with this detailed of a rating system is that the
> influences involved either affect the score too much or too little, and
> that this will turn into a matter of individual taste.
>
> Of course, you could always build in a facility for people to create
> their own ranking algorithm and just supply a sensible default.

Well, if we can't find one, the average user will probably not be able to
find a better one.

Marc


Marc Battyani

unread,
Dec 8, 2005, 6:37:22 PM12/8/05
to

That one is easy to implement but won't work with URL which have some
session encoding in them.

Marc


Marc Battyani

unread,
Dec 8, 2005, 6:39:38 PM12/8/05
to

"Tayssir John Gabbour" <tayss...@yahoo.com> wrote

> Marc Battyani wrote:
> > > A new interface ? Other ideas ?
> >
> > I have seen that ther are already some posts with very interesting ideas
and
> > propositions. I can't really reply to them now as I somehow have to do
real
> > work. ;-)
> > But you can continue to propose ideas anyway. Also you can post code
> > snipsets to support your ideas. In Common Lisp obviously, or in
javascript
> > for the interface.
> >
> > I will reply to them this evening. For now I will just check it from
time to
> > time to see if all is ok.
>
> Some things I've noticed as a user; ignore any which are mistaken:
> * Adding categories is very slow... (Maybe you're intending this to be
> some wizzy javascript thing later, but right now giving them its own
> page?)

I will check that.

> * Nice just to submit urls without going to a special page.

???

> * I'd like to stay logged in between browser visits.

OK that one is easy.

Marc


Marc Battyani

unread,
Dec 8, 2005, 6:46:30 PM12/8/05
to

<pkh...@gmail.com> wrote

> Marc Battyani wrote:
> > I have seen that ther are already some posts with very interesting ideas
and
> > propositions. I can't really reply to them now as I somehow have to do
real
> > work. ;-)
> > But you can continue to propose ideas anyway. Also you can post code
> > snipsets to support your ideas. In Common Lisp obviously, or in
javascript
> > for the interface.
>
> First, wow. I can't express the amazement at the amount of work you're
> managed to do in only a couple hours.

Thanks ;-)

> Second, my suggestion: Ideally, we'd want to use scoring information by
> each user to tailor their link list, and show them links they are
> likely to appreciate (probably because a group of similar users liked
> it too). In other words, give greater credence to the scoring of user
> who score other articles similarly as us. However, with only an
> unidimensional scoring system, extracting meaningful information is
> obviously harder than with a more expressive scoring system. So, how
> about scoring with a tag and an up/down arrow? Something that allows me
> to express "tag: Good-Lisp-Language; score: down", meaning that it's
> about (good) Lisp and I don't like it [implying that it's because it is
> an element of Good-Lisp-Language]. Note that the tag system allows to
> decouple preferences and judgement of quality. It might be better to
> actually have the ability to score both the article and the relevance
> of the tag separately. With this system, one could only vote a tag up,
> not down.

Seems a good idea to rate individually the coolness factor by categories.
But I don't know if the users will want to give 5 appreciations for one page
for instance. Maybe this can be computed from the statistical data. for
instance if a link with categories lisp and web are rated good by people
having lisp in their favorite categories and bad by people with only the web
category than we can increase the lisp coolness and decreas the web one.

> Finally, the eternal problem with tags is that we end up with tons of
> irregularly-named, unclear, redundant or overlapping tags. The clarity
> and regularity of the names should probably only be enforced by having
> a style guide for tags, and not using tags of bad style. As for
> redundance and overlapping, some dynamism and human laziness could
> help. When scoring an article, a popup (yet more JS :/) could appear,
> allowing one to choose popular (in general, or for that entry) tags, or
> to enter one in a form (here, something like Google Suggests could help
> alleviate further the problem of bad tagging).

For now, the categories are fixed by the administrators.

Marc


Marc Battyani

unread,
Dec 8, 2005, 6:49:17 PM12/8/05
to

"Marc Battyani" <Marc.B...@fractalconcept.com> wrote

> The second release of linkit is online at the usual place:
> http://linkit.fractalconcept.com
> Be sure to register to see the new functionalities like the positive and
> negative filters for categories and submitters and the ajax like stuff.
>
> It is what I have so far after about 7h of work.

Well I couldn't find the time to work on it again today. So I just made the
categories clickables in the link lists.
It's nothing but I find it cool ;-)

Marc


Marc Battyani

unread,
Dec 8, 2005, 6:51:33 PM12/8/05
to

"Paul Wallich" <p...@panix.com> wrote

> Emre Sevinc wrote:
> > Providing users a few ranking schemes (with fancy names?) from
> > a drop down list, and also giving them the opportunity
> > to create a brand new ranking scheme, give it a name (and
> > an option to make it public) and save it. This sounds very cool
> > and innovative.
>
> That would be particularly sweet if you let users switch between schemes
> easily (e.g a bunch of url suffixes so that they could see different
> slices of the site depending on their mood or interest of the moment.
> (Sometimes I want to see a focused set of stuff, sometimes not.) Then
> you could go meta and watch how many people used each ranking scheme,
> and offer them to other people based on that...

Good idea: switchable preset filters.
Going meta is the way to go! ;-)

Marc


Marc Battyani

unread,
Dec 8, 2005, 6:58:48 PM12/8/05
to

"Matthias" <n...@spam.please> wrote

> Depends on what point you want to make. If you want to make it cooler
> than the competition the next natural step is to improve usability.
> These are seemingly simple things such as not reloading the complete
> page when a user rates a link. Or removing the http:// entry in the
> submit form. And encouraging the user to give _short_ descriptions
> for their links. (Maybe "title" should be "one line description" or
> so, the "title" can be inferred from the web page if needed at all.)

There is already a desciption field in the submit page but it's not yet in
the links list as reddit has not put it (digg does though)
Maybe I will put this as an option for the users

> It's very impressive how fast you got the infrastructure working. But
> I don't think that the infrastructure is the hard part here. The hard
> part is the social engineering. (And that's not just the ranking
> function.)

Yes we all agree there. It's now that the fun begins

> Concerning comments and all sorts of cool features: I would introduce
> them very carefully, only when I'm almost positive they are a good
> idea. I'm quite sure that you are capable of building more features
> faster than most people. Unfortunately, this alone won't create a
> cool site.

Well as the only users are developpers, we can try thing without loosing too
much users.

> Finally, I can't but also give a recommendation for the ranking
> function: :-) My very first try would be to sort links according to
> how often they were clicked in the last couple of hours. I wouldn't
> mind if they are ranked up or down or who clicks them. Just count
> clicks. And also: Don't tell users what, precisely, the ranking
> function looks like. People like small miracles. :-)

Yes good idea, this couls at least be counted as a interest factor to use in
the ranking function.
(correlated with the categories for instance)

Marc


Surendra Singhi

unread,
Dec 8, 2005, 8:30:33 PM12/8/05
to
"Marc Battyani" <Marc.B...@fractalconcept.com> writes:
>
> I have seen that ther are already some posts with very interesting ideas and
> propositions. I can't really reply to them now as I somehow have to do real
> work. ;-)
> But you can continue to propose ideas anyway. Also you can post code
> snipsets to support your ideas. In Common Lisp obviously, or in javascript
> for the interface.
>
Hello,

Is the lisp code for the website available somewhere?


Thanks.

--
Surendra Singhi
http://www.public.asu.edu/~sksinghi/index.html

,----
| Great wits are sure to madness near allied,
| And thin partitions do their bounds divide.
|
| (John Dryden, Absalom and Achitophel, 1681)
`----

Julian Stecklina

unread,
Dec 9, 2005, 5:32:41 AM12/9/05
to
Creighton Hogg <wch...@login01.hep.wisc.edu> writes:

It would be quite difficult not to make this a gaping security hole or
a totally crippled domain-specific language. Has anyone done something
like this before?


Regards,
--
Julian Stecklina

When someone says "I want a programming language in which I
need only say what I wish done," give him a lollipop. - Alan Perlis

Marc Battyani

unread,
Dec 9, 2005, 6:15:46 AM12/9/05
to

"Surendra Singhi" <efuz...@netscape.net> wrote

>
> Is the lisp code for the website available somewhere?

Yes, I've put it on linkit: http://linkit.fractalconcept.com
(I cheated and set its number of points to 100 ;-)

Some comments:

-It has been written in 7 hours so it's not optimal nor polished.

-Most of the time has been spent figuring out what to do rather than coding.

-It's deceptively short (260 lines) and boring.

-All the database and web stuff is handled by the framework ;-)

-The definitions of the object classes are done in the Framework through the
Framework UI

-Some functions look like they were written at 3AM ;-)

Marc


Surendra Singhi

unread,
Dec 9, 2005, 8:42:12 PM12/9/05
to
"Marc Battyani" <Marc.B...@fractalconcept.com> writes:

> "Surendra Singhi" <efuz...@netscape.net> wrote
>>
>> Is the lisp code for the website available somewhere?
>
> Yes, I've put it on linkit: http://linkit.fractalconcept.com
> (I cheated and set its number of points to 100 ;-)
>

Thanks, I was planning to write some web application and I was struggling with
Joomla and PHP hell.
It looks great and even more incentive to learn the framework.

Thanks again.

,----
| "War is Peace! Freedom is Slavery! Ignorance is Strength!"
| -- Orwell, 1984, 1948
`----

Rob Warnock

unread,
Dec 10, 2005, 8:24:59 PM12/10/05
to
Marc Battyani <Marc.B...@fractalconcept.com> wrote:
+---------------
| > I [will] correct this.
|
| Corrected.
| People doing web development without a listener are missing something great!
+---------------

I second that opinion! It is *so* great! A source editor, a browser, and
a REPL window -- what else could one want?!? [O.k., a cup of hot coffee...]

Plus, if you turn on logging [e.g., I run my web app servers under
"detatchtty" <http://www.cliki.net/detachtty> with a "dribble file"],
then even if you don't have an active connection to the REPL at the
moment something goes wrong, you can often look at the log file and
get enough information to get your debugging started in the right
direction when you *do* (re)connect.


-Rob

-----
Rob Warnock <rp...@rpw3.org>
627 26th Avenue <URL:http://rpw3.org/>
San Mateo, CA 94403 (650)572-2607

Bernd Schmitt

unread,
Dec 12, 2005, 4:14:16 PM12/12/05
to
Thank you very much

for doing this interesting lisp-online-app.
It would be great, if you oould put this on a public repository. I would
suggest gna.org, because they do respond *very* fast to user problems.

If you like i can put it up for you (with your copyright, a licence of
your choice etc...)


Ciao,
Bernd

--
T_a_k_e__c_a_r_e__o_f__y_o_u_r__R_I_G_H_T_S.
P_r_e_v_e_n_t__L_O_G_I_C--P_A_T_E_N_T_S
http://www.ffii.org, http://www.nosoftwarepatents.org

Tayssir John Gabbour

unread,
Dec 13, 2005, 7:12:03 PM12/13/05
to
Bernd Schmitt wrote:
> Thank you very much
>
> for doing this interesting lisp-online-app.
> It would be great, if you oould put this on a public repository. I would
> suggest gna.org, because they do respond *very* fast to user problems.
>
> If you like i can put it up for you (with your copyright, a licence of
> your choice etc...)

Is this likely? It appears Linkit was created with proprietary tools --
one that Reddit didn't have access to. So in that case, impressive as
Linkit is, I'm not sure I'd use it to claim Reddit's decision was
flawed.
http://linkit.fractalconcept.com/asp/70e9/sdataQkvZcw9ozyZLDQjr-br28uLk8yBX8yBX8oQ5Ssjh-15rDM==/sdataQu3F$sSHnB==

But perhaps I misunderstand the situation.

Tayssir

Surendra Singhi

unread,
Dec 13, 2005, 8:19:54 PM12/13/05
to
"Tayssir John Gabbour" <tayss...@yahoo.com> writes:
> Is this likely? It appears Linkit was created with proprietary tools --
> one that Reddit didn't have access to. So in that case, impressive as
> Linkit is, I'm not sure I'd use it to claim Reddit's decision was
> flawed.
> http://linkit.fractalconcept.com/asp/70e9/sdataQkvZcw9ozyZLDQjr-br28uLk8yBX8yBX8oQ5Ssjh-15rDM==/sdataQu3F$sSHnB==
>

I think this is some kind of bug in linkit. I used the above link posted by
Tayssir to go to linkit website, and the website started thinking of me as
'Tayssir'. I posted some comments, but they got posted under Tayssir's name.


Cheers.

,----
| "All animals are equal, but some animals are more equal than others."
| -- Orwell, Animal Farm, 1945
`----

Surendra Singhi

unread,
Dec 13, 2005, 8:23:38 PM12/13/05
to
Surendra Singhi <efuz...@netscape.net> writes:

> "Tayssir John Gabbour" <tayss...@yahoo.com> writes:
>> Is this likely? It appears Linkit was created with proprietary tools --
>> one that Reddit didn't have access to. So in that case, impressive as
>> Linkit is, I'm not sure I'd use it to claim Reddit's decision was
>> flawed.
>> http://linkit.fractalconcept.com/asp/70e9/sdataQkvZcw9ozyZLDQjr-br28uLk8yBX8yBX8oQ5Ssjh-15rDM==/sdataQu3F$sSHnB==
>>
>
> I think this is some kind of bug in linkit. I used the above link posted by
> Tayssir to go to linkit website, and the website started thinking of me as
> 'Tayssir'. I posted some comments, but they got posted under Tayssir's name.
>

And whats worse is that, I am not able to find a link to logout. So, every time
I go to it, I am shown as 'Tayssir'.

,----
| By all means marry; if you get a good wife, you'll be happy. If you
| get a bad one, you'll become a philosopher.
| -- Socrates
`----

Reply all
Reply to author
Forward
0 new messages