Where are the readers?

Forum: LXer Meta ForumTotal Replies: 85
Author Content
sbergman27

Jul 01, 2006
7:05 AM EDT
This is something I've been meaning to ask for a while.

From the right sidebar:

"LXer is read by around 500,000 individuals each month,"

Where does that number come from?

I'm going to make a guess and say that about 25 - 50 people have posted here in the last month. Does that mean that the poster to reader ratio is between 1:10,000 and 1:20,000?

Doesn't that seem... well... odd?

Koriel

Jul 01, 2006
9:44 AM EDT
I can say im still here just not posting often as im extremely busy after having just moved from the UK to US and trying to get our house ready and my PC's only arrived yesterday after being in transit for 2 months. Hope to get back to posting normally in about another 2 months once im settled in.

Real life has a habit of interfering in Net life, which is a shame :)
dinotrac

Jul 01, 2006
10:49 AM EDT
Steve -

Interesting question.

All the more interesting because I know that 500,000 number came about under *ahem* a different regime. I am quite certain that some of the people posting now were not posting then. I wonder how many people who were posting then are not posting now?

For that matter... What is typical for a site like this?
sbergman27

Jul 01, 2006
10:56 AM EDT
Well, I was kind of assuming that the 500,000 number was an optimistic estimate. I mean... 50 people talking... and 499,950 people sitting around silent? It's just not natural.

Edit: I meant to say 51 and 499,949. I momentarily forgot Koriel. Sorry, Koriel.
hkwint

Jul 01, 2006
11:15 AM EDT
Quoting: Where does that number come from?


From the webalizer statisticts.

Statistics still show ~450k readers per month. Don could tell you more about it maybe, since I'm not that into our statistics.
jimf

Jul 01, 2006
12:26 PM EDT
> 499,950 people sitting around silent? It's just not natural.

Well, that would probably be all the hits on the site. That really doesn't say anything about the number who participate on the site or about those who lurk, or even just read an occasional article. Actually it's quite natural.
hkwint

Jul 01, 2006
1:34 PM EDT
It really is 15.000 (15,000 for US/UK) unique visits per day. A month has 30 days... You do the math!
dek

Jul 01, 2006
1:56 PM EDT
How do you separate the repeat visitors out? Or do you? I know I visit several times a day just to keep current on the discussions, Not to mention the RSS feed.
jimf

Jul 01, 2006
2:52 PM EDT
Probably the most accurate count of significance would be the number of members.
sbergman27

Jul 01, 2006
3:15 PM EDT
> Probably the most accurate count of significance would be the number of members.

Active members. Members that have logged on in, say, the last 30 days.
dinotrac

Jul 01, 2006
3:55 PM EDT
From the Hans's description, it sounds more like 15,000 people, give or take. I'm quite sure visitors don't do a once-a-month rotation around here.
dcparris

Jul 01, 2006
8:29 PM EDT
That number may be slightly skewed, as in March we had an extremely high volume of visitors. Since then, we have hovered around 450k unique visits. That may be the number Dave is using for the count. I believe that the "Unique Visits" count is based on the number of pages viewed in a given time frame. How Dave has that configured, I don't know.

If he sets that time frame for one hour, then every time you allow an hour to pass before checking the site, you trigger a new visit. I am not certain how RSS readers fit into the picture, but assume someone has to actually open an article from the reader. At any rate, if you visit the site frequently, you're probably being counted each time.

We actually grew tremendously from October through March. You can compare us to other news sites on Alexa to see where we stand. This shows where we stand in comparison to Linux Today: http://www.alexa.com/data/details/traffic_details?compare_si...

April and May saw a decline, but our June stats show us regaining ground. Internally, we look at the unique visits, page views, and unique referrers. Dave has the figures for our registered users. Historically, though, I believe the ratio of active to inactive readers has been pretty low.
jdixon

Jul 01, 2006
8:51 PM EDT
> Does that mean that the poster to reader ratio is between 1:10,000 and 1:20,000?

I've heard from normally reliable sources that the poster to reader ratio at most sites is in the range of 100:1. I don't think 1000:1 would be totally unreasonable. Any higher than that, and I would probably start questioning my numbers.
stevem

Jul 11, 2006
1:40 AM EDT
I may be able to help shed some light here - web analytics is both hobby (I have 3 GPL'd projects in that space) and part of my full time job. :-)

There is a *big* difference between visits and people. A visit is typically defined as "something" accesses a site. With no activity for > 30 minutes, any new activity counts as a new visit. Visits != People. A spider, eg GoogleBot, will count as a visit in normal log analysis unless to go to a lot of effort to filter most of them out. gawk is very useful here. :-)

It is normal for spiders to account for quite a large percentage (like 20 - 50% - depending on site) of page views, but usually do not impact hugely on "visit" numbers - mainly as they only come from a very small number of IP addresses, and usually come somewhat in a single burst.

You can get a better idea of actual people if you use cookies (a la apache's mod_usertrack) or page tagging (a la Google Analytics). Keeping in mind that Apache allows you to log all sorts of goodies above and beyond what a normal "combined" log gives you.

Webalizer uses IP address to determine a "person". This is wrong in both directions. As a trend tho it's generally pretty good - can see if you went up from week to week, or down. Corporate firewalls with NATing may obscure thousands of people to a single IP address. Proxies, particulary AOL's go the other way - a single person coming from multiple IP addresses.

Alexa and related technologies are so fraught with if's but's and maybe's that I would only use their numbers if /dev/random was broken. :-) As a comparison to identical style sites/competitors tho - it can be useful - the comparison the LT would be mostly valid. More as a shape of trends than percentage above or below. I've had access to a Hitwise subscription in the past (Whee - $20K per year...) and even those are pretty ordinary to use as a tool - for similar reasons.

Back to the original problem: Measuring people numbers is *hard*. If not impossible. You can approximate and get close and maybe say "+/- 5%", but it will still have an error component. eg. I use firefox, Konq and Opera on this PC when booted to Suse10.1 esle FFox, IE and Opera under XP-Pro; Installed Kubuntu, 6.06 not really used yet. FFox & Opera on my main W/S @ work (Suse 10.0); IE and FFox on the other PC. So how many people am I? :-) Hmm. But my wife also uses FFox on my PC under linux or winxp. Our 3yo just wants to close any open window, so browsing isn't an issue. Yet. :-) Aside - he really liked tuxpaint. Until I bought a Winnie the Pooh edu game (win only- natch...) for his b'day. I was terribly disheartend the day he burst into tears when I didn't stop grub in time from booting suse 10 instead of manually changing to windows xp and booting that. I've now, with heavy heart, changed the default grub boot choice to XP. Sigh. :-(

The flip side is that folk who use as many machines as me (and there will be others who use more!) aren't that common. But there's enough with work vs home use to help skew numbers. So the concept generally used is to accept the error, but do you best to get accurate numbers. If you can force people to login, you can get a very accurate idea of numbers - but even that can be subject to abuse.

If you only have access to webalizer style stats - look more at the "Site" numbers than the "Visit" numbers to get a better idea of # of people. And then ~ +/- 15% around that number to get a really vague idea of error. But without some sort of cookie tracking etc, even that 15% number is a total WAG.

HTH? Cheers!

Steve
dinotrac

Jul 11, 2006
3:03 AM EDT
stevem -

Yes, and it is compounded by a statistical fallacy: Unique visits per day * days in month = people in a month.

This kind of calculation is very common for things like festivals, sports teams, movie theatres, etc. To them, each ticket walking through a door is a person. That's true, but each ticket walking through the door is not a unique person across days/shows/events.

A sports fan may go to all of a team's home games - in baseball that would be 81 "people" for one flesh 'n blood being.

Likewise, lxer is multiplying the daily count by 30 to get a monthly figure. The total certainly does not represent unique people. Lots of folks like lxer enough to come back again and again.



stevem

Jul 11, 2006
6:36 PM EDT
@dino

I'm not so sure there is a compound problem. webalizer resets daily numbers ... daily :-), but the monthlies are kept separate and only reset when a new month rolls around. So if the reliance is on the monthly totals, the compounding isn't such a big issue. As in, the internal hash structures in webalizer maintain the individual IP address/name mappings for a month. You can easily see this if you look at the .current file - which maintains state between (for eg) daily runs.

It'll still be wrong, but not as wrong as 15,000 x 31 for example. There is more of a dependance on people having static vs dynamic ip addresses - which is usually broadband vs dialup.

Hans implies the 15 x 31, but I suspect he was looking at the daily average column. That's a sneaky column if you have traffic that has major swings during a week. One site I manage has a huge downturn on weekends - like nearly half our M-F traffic. That has a huge negative impact on averages. And is, hence, quite misleading.

Or am I missing something? :-)

To the site owners and team: Don mentions referrals as a stat to track. The best - and I do mean best - referral analysis tool I have *ever* come across is a GPL'd perl program: http://ktmatu.com/software/relax/

I am comparing with Google Analytics; Webtrends ; SAS's webhound and several others. Some are $100K a year plus products. Relax really *is* that good.

What I found interesting looking at the Alexa graph Don linked to was two things: 1. There was a huge upswing in Late April '06 followed by a big drop immediately after. 2. Despite matching it with LT late last year, and early 06, things have fallen away since.

So questions which arise: a. Is that upswing significant. Tied to a single event or??? b. Has there been a general downward trend? Or is it more of an Alexa randomising issue? Why/Not?

I'm not suggesting the questions be answered here :-) rather these are the starting broad brush questions I'd be asking and trying to find the answers for. With the answers you can then start to implement change (or not) to fix or improve things (or not) as appropriate.

Fascinatin' stuff. :-)

Cheers!

Steve
grouch

Jul 11, 2006
7:29 PM EDT
dinotrac:

The webalizer FAQ states that it does not use daily * days-in-month for monthly stats.

Regarding stevem's comment, "Site" numbers show the same spike in March as "Hits" and "Visits", (remember Tom's article that stirred the hornets of digg and Slashdot?) with a decline back to a level a little higher than it was in Oct, 2005. I'm still trying to make sense of the stats. Average daily "Hits" are about 7 times avg. daily "Visits" and avg. daily "Sites" are about 2.5 times avg. daily "Hits". I've given up trying to figure out how many people are reading and just watch the bar graph to see if it fluctuates a great deal from month to month.
dinotrac

Jul 11, 2006
8:23 PM EDT
grouch and stevem:

Well that's a relief. Thanks for setting me straight.

Better something that's honestly flawed than just plain wacky.
wjl

Jul 11, 2006
9:51 PM EDT
Hi all, even if we cannot trust the numbers in Webalizer et al, these are nonetheless quite interesting as some of you said already. My personal homepage climbed by the factor of 30 from last September, since I set up a blog and write about mostly what I'm doing at work and at home, using only Linux and FOSS. Still my number of visitors/sites are also by the factor of 30 lower than the ones on LXer for instance, but although I'm making "progress" in that regard, I'm much more interested in the "search term" columns on my site. Amazing, how many people for instance want to know and learn about Xen now for instance. Anyway: sure the Webalizer and its colleagues could use some kind of improvement, but since I neither have the time nor maybe the knowledge to do it, I take these numbers as they are - only to watch and compare monthly progress (of course my site is doing much better than the commercial one of my employer ;-) ). Thanks for the links and for this discussion - oh and thanks to the LXer staff of course for all of their work...
dcparris

Jul 11, 2006
10:57 PM EDT
O.k., I came on board around the end of October. Throughout November, we had Tom and Carla at the helm. We also had Paul Ferris and a few other editors - and we were all writing fairly regularly. Carla resigned at the end of November, which was then our peak month. We managed to maintain a respectable level during December, January and February, despite the loss of Carla and a few other editors. Even I got busy with personal issues during that time.

We had a surge in mid-March, tied to a single article written by Tom. Tom left in April and it took me until May to recruit grouch. We reached a low point in May, albeit still higher than Dec-Feb. We have been climbing back up since then. We are currently coming much closer to LT again. I anticipate we will surpass 500k unique visits this month - probably with some ease. I expect that trend to continue upward, even if somewhat slowly.

With respect to the Webalizer stats, visits and page views don't always correspond. Thus, even though our visits were up last month, our page views were down, as compared to May (the low month of the 2nd quarter). Likewise, the referers stat is the same way - it doesn't correspond to much of anything, but seeing if 2000 sites refer us in June, and 4000 in July, then we know that even though some of those are dupes from June, more sites referred us than before.

This thread got me interested in comparing our registered users to the overall stats. Here's the thing. We all know by now about "lies, damn lies and statistics". But what we have does give us an idea of what we're accomplishing each month. In the Summary page, the "daily average" includes the carry-over at midnight, which skews the actual average.

I send out a monthly e-mail analyzing where we stand, based on the stats. I realize they are not the most accurate reflection of what's going on. Even so, I'm learning a great deal about (and from) the stats. My thinking is that it has taken me 2 or 3 months to get my footing as an editor-in-chief. :-) Recruiting grouch has proven to be a smart move, even if no one ever believes that. ;-) I'm working to re-invigorate the editorial team, generally.

If we proceed from here with a robust story-posting cycle, increase the number and regularity of original articles, and continue building the sense of community I see developing, we should continue to see steady growth from here. How does that strike you guys?
wjl

Jul 11, 2006
11:27 PM EDT
"If we proceed from here with a robust story-posting cycle, increase the number and regularity of original articles, and continue building the sense of community I see developing, we should continue to see steady growth from here. How does that strike you guys?"

Sounds like a plan... ;-) No, really: keep up the good work. I have to admit that I like Tom and what he's writing, but still... this is by far the best Linux News site except maybe some of the really old ones like lwn.net and such. For daily (or even hourly) news, there is nothing like LXer. And that pretty much explains the Webalizer stats IMHO.
grouch

Jul 11, 2006
11:49 PM EDT
>"Recruiting grouch has proven to be a smart move, even if no one ever believes that. ;-)"

The fine print on the box said, "Sublim-O-Message does not guarantee results. Not responsible for hives, twitches, mental lapses, nor reduced conciousness."

To paraphrase the great Bart Simpson, it's not my fault and you can't prove a thing.
hkwint

Jul 12, 2006
12:44 AM EDT
Quoting:Hans implies the 15 x 31, but I suspect he was looking at the daily average column.


Yes, right. But I warned I wasn't that up to date with our stats, at least I hope I did. It seemed to logically to be false, since it matched real closely.

BTW it really surprises me that it's almost impossible to count people on a well-monitored thing like the internet. You' d say it was easier than counting people on a festival or so, but after reading this thread, I think it may be the other way around.
grouch

Jul 12, 2006
3:14 AM EDT
hkwint: >"BTW it really surprises me that it's almost impossible to count people on a well-monitored thing like the internet."

stevem's comments were particularly enlightening, for me.

(I never know whether 'tis more proper to capitalize when beginning a sentence with a lower-case pseudonymn or to keep the choice of the owner of the 'nymn).

We need a mod_head_counter for Apache. Or at least some bit of hacked code that could count fingers on the keyboard and divide appropriately. Oops, that leaves out the voice and puff users. (Would also omit dinotrac, who is known to pound vigorously upon his keyboard with a club or small defenseless animal).
dinotrac

Jul 12, 2006
3:34 AM EDT
grouch:

Hey! That Chihuahua ain't defenseless!!!

His yelps are very unpleasant on the ears. Headphones help.
stevem

Jul 12, 2006
3:48 AM EDT
Goodness, there'sa bit to reply to now! So much for an early night! :-)

@wjl: It's not that webalizer can't be trusted. And I do apologise if I gave that impression. Webalizer is extremely accurate. Its the interpretation of what it's telling us that is broken. When it says visits or sites, we tend to assume people. It's not. Spiders, IIS worms, Monitoring..., the list is huge of what can register against a site. The trick is to strip the gumpf out of the numbers and reduce down to what's important. And trust me, if you cut your numbers to only a fifth you're doing *really* well. :-)

I'll use LXer as an example of what I'd focus on: The main goal with this site, as far as I can tell :-), is two fold. 1. Revenue from Ads - which pays people and infrastructure. 2. All things Open Saucy (Sorry couldn't resist the bad pun). How do you measure these? Well 1 at first blush is easy. Google does it for 'em. But #1 is heavily influenced by #2, so I'd focus on #2.

So look into #2 in more detail. What sort of success measure is there? How about: How many links off LXer does a given person follow? The current setup makes that pretty easy to measure. A simple cut | sort |uniq -c | sort -rn style analysis can help there. That would be the starting point. Next would be to drill more out of the logs. What categories of articles work? What doesn't? Possibly why/not? Is it worth getting Tom to write more articles to inflame Digg and /. users? ;-) Hmm that last is more editorial discretion I suspect. :-)

Just cause something doesn't work doesn't mean it should be dropped. It may add value in a more undefined way. Say "trust", or help affirm a statement - eg a link to spreadfirefox. That in turn can add credence to a site even if not directly useful. "I'm hip with you guys" if you ken. Articles of faith, preaching to the choir. etc.

But again the numbers can mislead - the obvious answer to a goal of get more people to follow more links is to get more links up! Right? Right! Wrong. At risk of starting controversy, one thing I find LT does really well is that it's volume of articles is quite tolerable. And they are very focused. Tho I *really* wish they'd stop linking to sys-con sites (a. The whole MOG/PJ thing still offends me, & b. I find the syscon sites horribly replusive to read. Garish and just plain yuk).

So the trick now is to come up with ideas and try 'em out. The good thing is you can get near instant feedback as to what does/doesn't work. Like days time frame. If you're clever about it, you don't have to do an entire site thing. Say only 1 in 5 users see's a new feature. Now measure who used it. Is it useful? Y/N. FWIW this is know as A/B testing. Amazon are noted for doing it continuously.

---

Don, be *very* careful with referral numbers - raw as in. There are bots out there that deliberately access sites looking like real browsers, but are solely trying to pollute the referral field. This is usually targetted at blogs rather than sites like this, but it will skew your numbers. I'd estimate of the 4-5,000 non search engine referrals we get a week, maybe 5-10% is this sort of pollution. YMMV. You can spot them cause they typically come from casino or "medical" sites. Spam basically. They also don't access any images.

As for visits and page views? No they probably don't relate too well. Spiders etc. Pages will be heavily inflated, usually. Compare the percentage of pages to hits (page/hit * 100 from memory. It's too late to open eclipse and find out. :-) ). This can give an idea of spidering activity. High %? Lotsa spiders. Low %, not so many spiders.

On the not-mentioned site I reference above, we get around 25% of all page counts done by spiders et al. And that's excluding all my monitoring and such already. We average around 40-50,000 pages a day to give you a compare/extrapolation point.

Midnight carry over? As in you rotate at (say) 2am? Ahh that'd be painful. We use cronolog and consequently our logs are purely daily. Well that's not quite true. Some of the slower pages can run into the next day, but they get "merged" back as having happened at the previous log entry's time. ie Monday:0:00:03 instead of Sunday:23:59:56. If you can't use cronolog, suggest a mix of 'date --date="yesterday"', mergelog and gawk or ?grep to get you out of trouble. Use as a prefilter to webalizer.

Hmm.Just re-read your last line. Didn't mean to imply above that pushing out more articles is a bad thing - just that it may not be a good thing. It Depends. :-)

Arggh, My fingers are going to fall off. Too much typing. Blood gumming up keyboard.

BTW if any of you are *really* keen, and fancy a trip to Canberra Australia, I'm doing a half day tutorial on this stuff in the last week of July at a user group style conference? ;-)

- Steve
stevem

Jul 12, 2006
3:59 AM EDT
> His yelps are very unpleasant on the ears. Headphones help.

:-) Jokes aside, I found that using my high quality powertool ear muffler protector headphone thingies *really* useful for coping with those 2am nights with a new baby suffering a bad case of colic. They took just enough of the "edge" out of the crying to help ones sanity remain. I pass this experience onto all would be new parents. :-) I wonder if I'll be able to do the same when he's a teenager....

Grouch? I know your pain. I prefer Steve, but will know who you mean, and won't take offense whatever you use. I personally am sufficiently old fashioned that using Nym's grates like fingers down a blackboard. Much prefer a first name. Maybe that makes me weird in an online world? ;-)

I suspect I do need to get off my butt and pull these "emails" and such I've been doing for several years about web stats into real articles. One day....

Steve
grouch

Jul 12, 2006
4:53 AM EDT
Steve:

I would read that article, especially if it was a feature on LXer. (I plug without shame!)
dcparris

Jul 12, 2006
7:51 AM EDT
Stevem: > Is it worth getting Tom to write more articles to inflame Digg and /. users?

You may have missed this bit above: Tom left in April and it took me until May to recruit grouch. As in Tom is no longer part of the team. I've seen some of his articles on LJ, and I don't mind linking to them. Losing Carla, and then Tom were tough losses to be sure. Part of my point, though, is that we seem to be doing quite well despite the adversity. This month will be the second best since our March peak - without having written any major home-grown articles.

To be honest, I have lots to learn about being an Editor-in-Chief. For that matter, I have lots to learn about the news business, period. Not that my lack of experience has ever stopped me before - I always jump in over my head. ;-) Positive feedback from our core readers, as well as from Dave, with respect to the general operation of the site, how I've managed forum disputes, and other things has helped me grow more comfortable with this position. Getting good ideas from our editors has been very helpful to the site as well.

As for the stats, I prefer AW Stats over Webalizer, though I believe you can customize both to some extent. AW Stats separates out the bots & spiders for you (or at least shows what percent of your overall stats are bots and spiders). I failed to mention that we also get to see just how many times each story gets viewed, regardless of whether it's an internal or external article.
jdixon

Jul 12, 2006
7:58 AM EDT
> and it took me until May to recruit grouch.

It's OK DC, everyone's allowed one mistake. :)
dcparris

Jul 12, 2006
8:08 AM EDT
Jdixon: It's OK DC, everyone's allowed one mistake. :)

You are so forgiving. :-)

Stevem: I need to second grouch's motion. Our readers could benefit from a solid article or two, featured right here on LXer. Just use that "Publish it here" link you see in the side box. And you don't have to stop at just one. Well, maybe after the first one we'll change our minds, but please do take a stab at it. ;-)
tuxchick2

Jul 12, 2006
8:12 AM EDT
> Is it worth getting Tom to write more articles to inflame Digg and /. users?

Not speaking of Tom in particular, who brought a lot of good energy to LXer, but as a general principle I think it's best to avoid that sort of thing. My preference is to attract a solid core of sensible, thoughtful readers. Slashdot, Digg, Jerry Springer- sure, you get eyeballs, but are they the kind you want? ZDNet is a prime example of anything-for-clicks. Their sole purpose is to pimp for advertisers, with predictable results- almost pure garbage. Their real customers are their advertisers, not their readers. David Berlind is an occasional bright light, and few others. But overall I wouldn't wrap fish in it.

Quality works, as LXer is proving. Good work, Don and Grouch and Hans and everyone.

dcparris

Jul 12, 2006
9:33 AM EDT
TC: I need to frame that and keep it where I can see it.
grouch

Jul 12, 2006
9:42 AM EDT
She just mentioned you out of politeness. I'm really the alabaster statue in her garden of journalistic admiration. Note that I am placed in the ~center~ of the list of names. That is irrefutable proof that...

No, wait, I don't want to go back into the little room yet! No windows, you say? Ok. That's different. Nice uniforms.
dcparris

Jul 12, 2006
10:00 AM EDT
> She just mentioned you out of politeness.

Um, I was referring to her content spiel. On the other hand, it's her business if she elevates me to journalistic demi-deity. ;-)

...squeezes head through doorway while exiting room.

Besides, she actually has you just left of center in the grand scheme of things.
NoDough

Jul 12, 2006
10:31 AM EDT
>Besides, she actually has you just left of center in the grand scheme of things. [emphasis mine]

Ummm. No political discussion allowed. ;-)
dcparris

Jul 12, 2006
11:22 AM EDT
We're a sick bunch.
stevem

Jul 12, 2006
2:27 PM EDT
Sick? Maybe. Highly entertaining? Very.

I must confess I've been reading lxer for a while but never really ventured into the forums - only occaisionally reading comments attached to articles. How's the quote go??? "I only read Slashdot for the articles" ??? ;-)

So all of the transitioning/changes have been invisible to me. Thanks for the heads up!



AWStats and Webalizer do slightly different things. I'll admit to a strong bias for webalizer if only that one of my projects is a major fork of same. :-) Seriously tho, AWS is more in the nature of tools like Google Analytics. Good for those who need to bounce numbers around a bit. But doesn't really give you the ability to really drill into the details. Webalizer (and family) are more of a reporting tool. Pretty. Produces a good overview so that you can then pull out the other tools and start digging.

My biggest gripe with AWS is that it is *really* slow. What I can do in my webalizer fork in under an hour (re-process a years worth of logs), takes around 15 to 16 hours in AWS. (~ 650 million lines of log) Multiply by 4 for a decent history and it gets *painful*. FWIW Webtrends is even worse.

The way I run webalizer is to produce dumps of various clusters of info, and post process those. Bring, say, a years worth together and then eyeball and regex filters/groups to add or remove. Wash, Rinse, Repeat. I have a fairly extensive gawk script to pull out the major bots and such that hit us. Prefilter, vs AWS's inbuilt filter. YMMV.

Grouch: You'd read the article where ever I put it. Except maybe zdnet. ;-) Shameless plug is fine. My /shameless plug tag was stripped. Should have used preview!

TC2 makes a great point - a clearly defined goal targetting a style of audience. So the trick is to now fiddle with the site and watch the numbers to see if that's what you're doing. Up or down. So ditch the pure "big" numbers, and try and find the actual numbers of that targetted group. No it won't be easy. :-)

* Should a pink hue be added to attract more ladies? :-) * Ditto blue for boys? * Does anything change if we move the nav bar from right to left? * What about the header title/about bar? Can we drop that to the page bottom instead?

Web sites are so cool in that you can make these fiddles on, even select parts of, your audience and get direct immediate feedback on how effective it is. They don't even have to email hate mail at you - you can observe how and where they're clicking.

I'll see what I can do about those articles... :-)

Steve
Scott_Ruecker

Jul 12, 2006
3:05 PM EDT
This is just my two cents worth..

I am as guilty as anyone for not contributing as much as I should or could as of late, that will not continue. But I just read this entire thread and I think we should quit worrying about stats that may or may not be accurate and focus our energy elsewhere.

It does not matter what the exact number of visitors is every day. Since I have 'been around' LXer, this site has done nothing but grow, in every single way. Yes the amount of traffic has gone up and down depending on the week you choose to look at the traffic stats, which are debatable. But everything else you can choose to look at is very positive.

We have made a big transition, lost and added new team members, increased the amount of original content, found ourselves on the front page of Google News and more I cannot think of right now. I may not deserve to feel this way because of my recent absence but I am Proud to be associated with this site and the people who are a part of it. I tell everyone I meet about it as I am sure you all do as well. Forget about the stats, this site is not going anywhere so it really does not matter if the traffic goes up and down because we have been on our way up for a long time and that's not going to change any time soon.

I'm glad I got that off my chest, those penny's were heavy. :-)

Scott
dcparris

Jul 12, 2006
5:10 PM EDT
stevem: > Didn't mean to imply above that pushing out more articles is a bad thing - just that it may not be a good thing. It Depends. :-)

I meant to mention this earlier. I think we have found our happy medium, in terms of story rotation. I just want to see us start producing more solid, original content. That has proven, historically speaking, to be the big draw. I'm working on that as I write this.

Scott: You have a point. We can't put too much stock into the stats. Still, it's nice to be able to measure our progress. One thing I can say, comparing our current stats to the last 12 months, we're currently working on our second biggest month since the March peak. And yes, you do have a right to be proud. All of us do.
Inhibit

Jul 13, 2006
10:07 AM EDT
sbergman27: In response to your statement about "is the ratio of posters to readers really that low". In answer, yes. PCBurn gets a decent draw of actual readers with very few posters, even on the original content.

'Course, most of that's straight reviews. But still, the ratio of readers to posters is a whole lot to almost nill. So the amount of posters here always seemed about right for the ratio to me.

And I've found AWstats to do a great job of stripping out the erronious entries from weblog analytics. And about a fifth of the traffic being "real" is right.. I usually segregate off two sections, one for "non-reader" and one for "RSS feed reader" traffic, in addition to the actual bonified people (or as close as I can get to knowing they're people).
dave

Jul 14, 2006
8:43 AM EDT
Interesting thread here; I've enjoyed all of it (including the jokes, which I've come to expect from this lively bunch)

LXer's webalizer stats are here:

http://lxer.com/411/

You need to be logged in to view them (stops spiders from indexing it, where referer-spam-bots can find it).

As another point, Dave's Garden (my other site) got 1,197,912 "Sites" (according to webalizer) and 16,830,891 page views (according to cut|sort|uniq) this past May. During that time, only 2,676 unique members posted to the forums. Then again, there's much more to that site than the forums.

Speaking of those numbers, if anyone wonders where I spend most of my time, they now know. :) Dave's Garden wasn't enormous when I started LXer but it is now and is what pays my bills.

Dave
grouch

Jul 14, 2006
9:29 AM EDT
Uh-oh. Busted. Cool it, folks, the cop's here.

May LXer grow big enough to support itself with just enough quality, unobtrusive, GNU/Linux-related advertisements. (Whatever happened to the cute Mad Tux?)

BTW, dave/Dave, I absconded with a few hits from your Garden when my significant other posted this on it: http://edge-op.org/grouch/plant_stand.html
dave

Jul 14, 2006
9:49 AM EDT
I'm all for good quality ads on the site; finding advertisers is a chicken-and-egg issue, though.

Mad Tux's campaign expired earlier and I did not pursue them for renewal...

BTW - Nice plant stand! I'm adding to my list of things to try this winter. If I get hurt, I know where to find you. :)

dave
sxf

Jul 15, 2006
8:41 AM EDT
re: lots of visitors, few comments I had started to write small comments here and there. Then I noticed that some comments got zapped without explanation, or became inaccessible (some by me, some by others) so I lost interest.

re: ads One of the major pluses of this site is direct linking to the articles (=! zdnet and many news sites deep into ad whoring, but = digg, /. etc). The flipside of this is that the average time spent on the site is low (correct me if I'm wrong), resulting in lukewarm interest by ad buyers.

This is why IMHO, it would be crucial for you to have a good volume of comments. People would check them, reply to them and therefore stay on the site longer. In addition, this will build (strengthen) the "site culture."

Ideas 1. transparent moderation (written rules for zapping + jailing of messages and user banning, identification of the mod who acted + reason, possibility of complaining and filing abuse etc) 2. comment voting + karma. This might be time-consuming (/ implementation or similar) but almost all the sites with a good comment volumes have done it. Look what happened over at OSnews... Note: this is a necessary condition, but not a sufficient one (e.g. technocrat)

Hope I did not across as too adversarial: I like the site. Thanks for the nice resource.

PS One wish: more technical articles, less advocacy: if the Zambian inland revenue service, the Portuguese ministry of education or the French police decide to switch to Linux is frankly irrelevant unless there are interesting titbits of technical info thrown in.
tuxchick2

Jul 15, 2006
9:04 AM EDT
"re: lots of visitors, few comments I had started to write small comments here and there. Then I noticed that some comments got zapped without explanation, or became inaccessible (some by me, some by others) so I lost interest."

I can't explain the inaccessible comments, but I did notice that back near the end of the reign of the previous editor-in-chief, reader comments were inexplicably zapped for no reason other than they disagreed with the EIC's viewpoint. I know this for a fact, as it happened to a friend of mine that I steered to LXer. I don't see this happening with the current editorial team.
sbergman27

Jul 15, 2006
9:09 AM EDT
My opinion about moderation is that moderation of posts should be a very rare thing. (And I believe that it *is* a very rare thing here.)

As to voting, if this site got to the point that we needed a karma system, I'd be looking for another site.
jimf

Jul 15, 2006
9:17 AM EDT
> I don't see this happening with the current editorial team.

agree. I had reservations when Don started, but, he's more than proven himself.

> As to voting, if this site got to the point that we needed a karma system, I'd be looking for another site.

Yep... Me too.

dinotrac

Jul 15, 2006
9:17 AM EDT
tc -

Yup. I know that I had stopped commenting altogether.

But not any more.

So...you see, it's not all roses.
grouch

Jul 15, 2006
9:41 AM EDT
If only Don would let me play with the big, shiny red buttons...

* on alternate Tuesdays, all posts would be piped through rev, * all comments posted before noon would get the rot13 treatment, * karma points would have a fixed, public price, * furst paust! would be auctioned, * off-topic comments about oatmeal would count toward valuable coupons and prizes, redeemable at favored blacklisted sites, * all comments disagreeing with indisputable facts of the universe (namely, disagreeing with my comments) would be rendered in 3 pt dingbat font.

We would double readership within a week!
sxf

Jul 15, 2006
10:58 AM EDT
Ok, ok, you are against comment voting... Is it because 1. it might be thought of as pandering to the vanity of posters or 2. the perverse effects (first post syndrome etc) or 3. it creates a "class system" of users?

1. is a moral/value judgement, so it can be countered only by putting forth another value judgement, which I do not want to do. I will only say voting increases comment volumes

2. the perverse effects are overwhelmingly a result of poor design (e.g. the +1 Funny at /. is stupid IMHO) or sometimes of the culture of the site (e.g. look at the comments that get upvoted at kuro5hin)

3. If you are against a class system consider that comment voting is separate from karma. Also, there are other systems such as Advogato's trust system (which was designed for diaries, but still - btw on the site there is an interesting paper on the subject)

In the end, if the comment volume does improve, a voting system will have to be introduced, as an increase in comment volume goes together with flamewars, attracts trolls...
sbergman27

Jul 15, 2006
11:05 AM EDT
Hmmm. I'm against it because LXer is, in a way, my family. And families don't need or want comment voting.

And I'm not sure I would consider a large increase in the number of comments to be an "improvement".

This isn't Slashdot. And I would never want to see it become Slashdot.
tuxchick2

Jul 15, 2006
11:07 AM EDT
frist post and hot grits to ye, grouch!
sbergman27

Jul 15, 2006
11:09 AM EDT
Yeah, and in Soviet Russia teh grits post YOU!
jimf

Jul 15, 2006
11:11 AM EDT
> In the end, if the comment volume does improve, a voting system will have to be introduced, as an increase in comment volume goes together with flamewars, attracts trolls...

Probably... I think Dave has to make a decision on how much growth is worth while.

I've observed there is a level of comment / growth where sense of community becomes muddled and lost. At which time a site becomes non viable for me. At that point, I'm gone, so voting is really a non issue.
sbergman27

Jul 15, 2006
11:18 AM EDT
> Probably... I think Dave has to make a decision on how much growth is worth while.

In unix-like OSes, we believe in small tools that do what they do well and work with other tools. LXer is a small tool that does what it does well and complements other sites.

BTW, on the topic of Slashdot, here is an automated Slashdot Story Generator for your idle amusement:

http://www.bbspot.com/toys/slashtitle/
Scott_Ruecker

Jul 15, 2006
5:55 PM EDT
A voting system? That would be great!

To be able to go to the voting booth and have my vote count and/or actually be counted would be awesome..

Oh wait, you mean a voting system on the threads here at LXer? Darn.

No, I think if we needed votes, we would need another site to go to in instead of LXer.

I have to admit that I like the auction thing grouch has going though..
jimf

Jul 15, 2006
6:09 PM EDT
> I have to admit that I like the auction thing grouch has going though..

Ah... Looks to me more like a yard sale :).
grouch

Jul 15, 2006
8:13 PM EDT
jimf:

You just want a cut of the vast profit potential.

-----

Moderation systems that rely on users ranking each other's comments always lead to a system that reflects the numbers of the group most determined to shape the system to suit themselves. Such a system will become a good illustration of "group think", but will not be fit for much else.

Personally, I'd rather rely on some person acting as moderator/dictator. That's what is in place here. When the moderator burns out and starts zapping every comment that uses fewer than x number of semi-colons, you gently pry the remote control from his talons and hand it to the next sucker.
jimf

Jul 15, 2006
9:24 PM EDT
> When the moderator burns out and starts zapping every comment that uses fewer than x number of semi-colons, you gently pry the remote control from his talons and hand it to the next sucker.

I'm not arguing, but, if Dave were to hand you the keys to the place, I wonder, how long would that be... You know that power corupts, and absolute power... well, I'd just hate to see you go down that road grouch.
grouch

Jul 15, 2006
9:47 PM EDT
jimf:

Ok, instead of having one Supreme Master Evil Overlord of the Universe and LXer, how about we share the power? You be the Grand Supreme Rajah of Gratuitous Graphics and Deceptive Decorations, and I'll be the Supreme Czar of Comment Corruption and Destruction. We'll fleece 'em for millions, selling protection from our activities!

[edited to add:]

jimf: >"[...] but, if Dave were to hand you the keys to the place [...]"

There may be some who read this thread who are unaware just how bizarre things can get around here. Let me assure those folks that Dave hasn't gone quite so insane as to hand me the "keys". Don has just been away for the weekend with his family, or something equally weird. Neither Dave nor Don have yet let me play with the really powerful buttons and knobs on the LXer bridge.
jimf

Jul 16, 2006
7:01 AM EDT
Well, I sure like the sound of millions. I suppose, as long as no animals are hurt in the process...

> Neither Dave nor Don have yet let me play with the really powerful buttons and knobs on the LXer bridge.

Heh...Those guys are just baiting you with that old carrot. Pie in the sky for the plebe ('watch him drool'). They gave that to me over at DebCentral. Them knobs and buttons seem awful fun for the first day or so, but, after that it's just more drudgery... Definitely not what it's chalked up to be.

dinotrac

Jul 16, 2006
7:59 AM EDT
>Them knobs and buttons seem awful fun for the first day or so, but, after that it's just more drudgery...

Well, there are two that have potential. The one marked "Wealth, Fame, and joy beyond your wildest dreams" could be good. As to the "Hideous end of all life as we know it" button, it doesn't sound good, but is at least interesting.
jimf

Jul 16, 2006
8:43 AM EDT
> The one marked "Wealth, Fame, and joy beyond your wildest dreams"

On a Linux related site, that usually dumps a huge pile of hearing on your plate.

> the "Hideous end of all life as we know it" button

Get out of the water. The Orca are coming.
grouch

Jul 16, 2006
8:58 AM EDT
Ok, if the guesstimated ratio of readers to active commenters 'way back up the thread is correct, there are over 1500 people who have read what we 3 have posted in this thread so far today. Would they be calling 911 or the DEA at this point?
dinotrac

Jul 16, 2006
9:45 AM EDT
grouch -

Neither. Too many proprietary systems, not to mention sharing with Big Brother.

Instead, they call and order a pizza. No pepperoni, please. It's out to get me.

Scott_Ruecker

Jul 16, 2006
10:05 AM EDT
OK, I've called the police and put out an APB on Don and Dave.

...if the suspects going by the names of dinotrac, jimf and grouch approach you, please remain calm and remember, at least one of them is known to be afraid of pepperoni, so be sure to have some with you at all times...

...LXer will return to its regularly scheduled programming after these messages...if Don or Dave receive this message, please hit the 'reset' button as soon as possible...

I couldn't help it. :-)
dinotrac

Jul 16, 2006
10:16 AM EDT
Scott...

The thought police came after me once, but left as soon as they realized that my alibi is perfect.
grouch

Jul 16, 2006
10:43 AM EDT
Please, somebody, for the sake of humanity, send hordes of pepperoni hurling in dinotrac's direction at high velocity!
dcparris

Jul 16, 2006
10:53 AM EDT
O.k., the Fifth Army has left Charlotte - or is still leaving anyway. My home was invaded by more than 50 people yesterday morning. Apparently, some of the troops got separated during the previous night's battle. My wife and I have begun the long process of sifting through the rubble of what was our glorious abode. Don't worry about us though. We'll just continue huddling in what's left of the hall closet for shelter til we can rebuild.

Meanwhile, I really appreciate the votes of confidence from my loyal subjects. I promise not to curse you with AOL CDs. And if I get really mad and hateful, I start shipping the Microsoft CDs. Of course, $300-$400 per CD might help to explain my seemingly unfailing love and patience.

I actually prefer the moderator model myself. Of course, the moderator's ring is, like the One ring, seductive and yet burdensome. It kind of plays on the original theme the editors were playing on when I first joined LXer. In the future, having multiple moderators, with some decisions being reserved for the group as a whole, would be far better than just giving grouch free reign with his army of snails.

I know it's not perfect, but it's one of the best approaches I'm aware of. The key is to find those who demonstrate good judgement and trustworthiness.
dinotrac

Jul 16, 2006
11:12 AM EDT
Hey Rev...

I agree with you re: moderator model.

Not everybody is cut out for the role, though -- myself included.

It takes a certain willingness to submerge your own ego while being willing -- only if needed -- to step in with a firm hand.

grouch

Jul 16, 2006
11:17 AM EDT
dinotrac:

Fits me perfectly: willingness to step in. Couldn't have said it better myself.

Strange to see a comment from you with only 4 words, though.
jimf

Jul 16, 2006
11:22 AM EDT
I've been in your position, and, agree with what you're saying Don. You're doing just fine here. grouch and I were just funning you... Well, at least I was, ya never can tell about grouch :).
tuxchick2

Jul 16, 2006
11:32 AM EDT
If you're seriously considering some sort of distributed moderating, then perhaps the TOS could use an overhaul. Specifically:

" Discussion and debate of a political or religious nature is not allowed on the site."

Like, get real. How much of the site content has elements of both? At least half?

"6. Do not place any material on our service that could be considered offensive, indecent, abusive, hateful, harassing, libelous, profane, vulgar or unlawful. Our audience tends to be professional in nature, and we have the right, but not the obligation, to remove, edit, or relocate any content that we feel violates the standards of our site. "

Lots of slippery slopes and gray areas, there. There is always someone eager to be offended by something. Where do you draw the line between "spirited debate" and "harassment"? "Vulgar"? Now really. Some of our best fun is, er, earthy. Abusive, hateful- all judgment calls.

It's good to have a team of mods, as the different personalities will balance each other and provide automatic checks. And cover different time zones.
jimf

Jul 16, 2006
11:57 AM EDT
> Lots of slippery slopes and gray areas

Exactly, and, that's what good mods are for. Someone has to make that final call, warn the offenders and take appropriate and (hopefully) fair action. Anyone who can't accept that doesn't live in the real world.
grouch

Jul 16, 2006
12:09 PM EDT
tuxchick2:

All you right-wing penguin worshippers are the same, constantly being TOS-Nazis just because you wear your combat boots too tight. Any sensible male will tell you that you only drop one wing when you're doing a barnyard dance. It takes both wings to fly straight.

(How'm I doing? Did I skip anything?)

I think "team" is the operative word, there. The automatic checks fail when the situation is more along the lines of a group of flunkies implementing a single, set policy. You've likely seen that kind of mess before, too.

I like the mushiness of Dave's TOS -- it incorporates and allows the judgement of people to be used in judging commentary.

There are some built-in checks already:

Dave is interested in having as many readers as possible, given the limits imposed by the site goal and target audience.

The target audience, being mostly composed of cats, wild dogs and donkeys, want to run wild in the streets in total, leaderless anarchy.

Don naturally wants a boot-camp of penguinistas, marching, saluting and digging foxholes with spoons, in preparation for the day we attack Redmond.

If Dave tries too hard for numbers, ala get-rich-click, the target audience votes with non-clicks.

If the target audience runs too wild, the /. effect takes over and the site becomes something else.

If Don drills too hard, people go AWOL, publisher notices, and we go 'round again.

A team of moderators, coached on when to "submerge [their] own ego[s]", and encouraged to "balance each other", and made aware of the character of the target audience, can exercise judgement to implement the TOS in a way that does not damage the personality of the site. (You can see such a thing in action at hotrodders.com, the largest such site on the Internet).

The worst problems come when moderators have ulterior motives, outside the TOS. That's when you need a chief to step in who has demonstrated integrity, and knows how to aim.
dcparris

Jul 16, 2006
2:41 PM EDT
> grouch and I were just funning you... Well, at least I was, ya never can tell about grouch :).

I know, forgive my feeble attempt at mixing humor with seriousness. :-)
stevem

Jul 18, 2006
2:00 AM EDT
Not that I think you need a vote, but I'd put my hand up for the moderation style system. It's the same sort of benevolent dictator model that many OSS projects work under in any event.

But then I'm biased. I have moderator privs (and then some...) on another very busy site. It never ceases to amaze me just how little actual moderating needs to be done once people get the idea that moderating is pretty much impartial and happens to everyone. I've been moderated there. Several times. :-)

It's a pretty thankless task tho, so my personal thanks and sympathy to anyone who does have that role.
dcparris

Jul 28, 2006
9:34 PM EDT
Update to this:

I just ran a little math out of curiosity. So far, we've had 459 registered users active on the site this month. That probably is not the total number of registered users. At any rate, if all 459 of us visited the site 10 times every single day of the month, it would only account for a little more than a quarter of the unique visits. 459 of us would have to visit the site over 32 times per day to reach the current level of unique visits.

Sorry, couldn't resist. :-)
dinotrac

Jul 29, 2006
3:05 AM EDT
But rev --

What if all 459 of us visited 10 times a day, but invited three friends to look at the screen each time we did?

Or...what if we visited 10 times a day, but were really, really intensely interested? Shouldn't that count 3 times as much?

Wait, wait!!!!

Shouldn't our inner child also count?

You know... That whole six degrees of separation thing......

I'm thinking that 459 registered users might actually be the same thing as millions and millions of visitors.

Or maybe just Kevin Bacon sightings.

I'm not sure. This stuff is all so hard.
sbergman27

Jul 29, 2006
6:29 AM EDT
> So far, we've had 459 registered users active on the site this month.

Better make that 460. We just gained Christophery.

Welcome to LXer, Christophery!
dek

Jul 29, 2006
7:31 AM EDT
> Welcome to LXer, Christophery!

Better read Christophery's postings before welcoming him! Looks like he's not too interested in discussion. However, making money via spam is more along his line. Sorry but that's kind of a turn off here.

Don K.
sbergman27

Jul 29, 2006
7:42 AM EDT
Now Don,

Christophery is keenly interested in getting LXers the best deals on cell phones and I respect that. It's not like anyone else around here has ever been so concerned with my cell-welfare.

-Steve
dinotrac

Jul 29, 2006
8:11 AM EDT
Steve -

Just so you know --

I care.

Hugs?
sbergman27

Jul 29, 2006
8:27 AM EDT
Dean,

At this late date, if you can't beat Christophery's prices, I'm not interested.

-Steve
dek

Jul 29, 2006
8:35 AM EDT
Let's all have a group hug!!

> Christophery is keenly interested in getting LXers the best deals on cell phones and I respect that. It's not like anyone else around here has ever been so concerned with my cell-welfare.

That's one way to take it, I suppose. Then, a logical extension of that is I can become fabulously wealthy buying QEGY.PK stocks that are going to just EXPLODE because "attune" cares enough to send me a email with the subject line of "tremendous arsenic"?



Don K.

Posting in this forum is limited to members of the group: [Editors, MEMBERS, SITEADMINS.]

Becoming a member of LXer is easy and free. Join Us!