| |
Mr. Mouse
Registered: Dec 2001
Posts: 235 |
CSDb - Quantify Me Too - Hype stopped?
Heya folks.
I thought I'd get the data of CSDb again for another analysis like I did for the CSDb - Quantify Me ( CSDb - Quantify Me ) release back in 2010. It's been now more than 4 years on, would be interesting to compare the pivotal line chart from that analysis with the current state of affairs:
In the chart you see the data of today, compared to the 31st of July 2010. You can also see a lot of more additions have been made to the database, most releases attributed to the period between 1984 and, say, 1997.
At that time in 2010, 88380 releases were in my analysis, of which 76425 did have a release year listed. Today, I counted 130743 releases, with 113612 having a release year. So that's 37186 new CSDb additions with a release year since back then. Of course, most of this is due to stuff that was apparently made back in the 80s and 90s.
I would like to draw your attention to the period after 2010 however, that showed a steady increase of yearly number of releases, from 1046 in 2010 to 2037 in 2013. The latter number is touching the number in 1984, mind you. In this light, I would point out the drunken rambling in a word document I put at the end of the third of these CSDb - Quantify Me pages ( http://www.xentax.com/?page_id=235, http://www.xentax.com/?page_id=254, http://www.xentax.com/?page_id=265 ). In effect, this one: http://c64.xentax.com/downloads/csdb_quantifyme_part2_mindspook..
There I suggested (so in 2010) that we might be reliving a hyped surge in C64 activity, not unlike that seen in the Golden Age of C64 (the 80s).
Further drunken analysis might now confirm that, since the rise in the years after that is pretty clear. ;-) Where's that bottle of wodka? Anyway, what is somewhat troubling though, is that 2014 seems to be a year of decline. Has the hype stopped, ladies and gentlemen? Have we all now gone through our mid-life crises and/or died and is there no future? ;-) |
|
| |
trent
Registered: Apr 2015
Posts: 12 |
Well I'm a year late I guess. Just started unboxing shit after 21 years of lugging it around with me year after year through one rental house after another after another after another then to our own home...
Nostalgia is a powerful thing - one can never truly escape it.
This stuff was such a big part of my life and development that to deny it is a disservice to what made me who I am today, and all that I've accomplished in the last two decades in life.
If my aunt hadn't given me that first C64 in 1986, and told me "learn to program", where would I be today?
Everything I have ever owned and every dime I've ever made is from programming, network management, server builds and administration.
The C64 is such an important aspect of our history as a species it helped shape the world as we know it.
Granted, it wasn't around for the long haul but it was *the* mainstream computing device for a long time, and several millions of people had their first exposure to computing on them.
PC's are ... well... they have no soul.
C64's and 128's, to this day, with all their quirks and what they are capable of doing with so little resources, still have a bit of magic to them. There's something just different enough, just unique enough, that you can't ever get out of a PC.
I don't remember much about my first {however many dozen} PC's and laptops over the years.
I still remember how the keys feel on my C128 and where I'd lay my cigarette down on the back when chatting with users on my BBS after they paged me.
You can't reproduce that feeling, excepting the one and only way to replicate it; to flip the power switch on the side.... |
| |
TheRyk
Registered: Mar 2009
Posts: 2070 |
Quoting Mr. Mouse... 2014 seems to be a year of decline. Has the hype stopped, ladies and gentlemen? Have we all now gone through our mid-life crises and/or died and is there no future? ;-)
Hard to tell :) Not sure about the hype being over. Hardware prices at ePay are still at an alltime peak.
Speaking only for myself, I'm not dead yet, "real" life is more critical than ever, project pipeline is well-jammed, and some coffins of 80s and 90s disks still wait to be transferred.
About statistics:
We could also say 2013 was - despite of no X party year - an unusual year of activity, e.g. +100 entries alone in PETSCII competition, numerous other rather well-attended CSDb standalone compos such as logo, music, font etc. Looking at 2015 so far (about 400 entries as of yet, BTW you can check out a great deal here:
http://csdb.dk/stats/stat.php#act ) about 1.600 might be what we can expect again, pretty much what we got in 2012 and 2014. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11114 |
one point in a graph like this does in no way imply a trend and is statistically irrelevant. nothing to see here, move on... |
| |
Mr. Mouse
Registered: Dec 2001
Posts: 235 |
Quote: one point in a graph like this does in no way imply a trend and is statistically irrelevant. nothing to see here, move on...
As if I don't know that, hence the winking Smiley. When will you ever learn to stop being such a prick. *rolleyes*
If anything, I eagerly await the 2015 data to show that the trend continues upward. :) |
| |
chatGPZ
Registered: Dec 2001
Posts: 11114 |
hit a sore spot eh?
i eagerly await this lame stats wanking to turn into actual release making for that matter. |
| |
Mr. Mouse
Registered: Dec 2001
Posts: 235 |
*chuckle* internet trolls never hit any sore spots.
But you are right about two things though: I do wank to statistics, having an epidemiological background, and I agree that releasing C64 software is preferable. :)
Nevertheless, stats can tell us a lot about the world, and even a little bit about CSDb and the dynamics of the releases these days. Since 2010, for two years in a row we saw a linear increase in the numbers (+400 each consecutive year), which is unprecedented since the year 2000. When, as is the case here, over a time span of about 15 years this never happened, then even a die-hard conservative epidemiologist will raise some eyebrows and ask "what's going on here?". And it just so happens that my analysis in 2010 was predictive of this. Be it done in a drunken state in a piece called mindspook. ;-)
And that in my opinion is the beauty of stats, of registries and of dirty data collections. Call it my hobby. I just like numbers. And I like to dig deep to find confounding factors. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11114 |
Quote:internet trolls never hit any sore spots.
just like statistics don't ever really tell us anything, except for what the one making them wants to tell us. |
| |
Mr. Mouse
Registered: Dec 2001
Posts: 235 |
Quote: Quote:internet trolls never hit any sore spots.
just like statistics don't ever really tell us anything, except for what the one making them wants to tell us.
I see. Well, in that case, I have some further reading that might interest you as well: http://content.time.com/time/specials/packages/article/0,28804,..
Back to topic, this is the distribution of the release types by release year, and it tells us the proportion of all the different types per year. Looks like especially large (relative) spikes were seen in graphics and music (and cracks) the past few years.
Or more specifically the past 5 years:
 |
| |
chatGPZ
Registered: Dec 2001
Posts: 11114 |
Quote:I see. Well, in that case, I have some further reading that might interest you as well
cool story bro. about as useful as this stats wanking (yet more entertaining) |
| |
hedning
Registered: Mar 2009
Posts: 4595 |
I like these kind of statistics, whatever they might say. Keep them coming. |
| |
MagerValp
Registered: Dec 2001
Posts: 1055 |
Quoting Mr. MouseBack to topic, this is the distribution of the release types by release year, and it tells us the proportion of all the different types per year. Looks like especially large (relative) spikes were seen in graphics and music (and cracks) the past few years.
Nice! Could you do a distribution graph with absolute values as well? |
| |
Oswald
Registered: Apr 2002
Posts: 5017 |
way too much releasetypes up there impossible to distuingish all those similar colors. |
| |
Mr. SID
Registered: Jan 2003
Posts: 421 |
Statistics like this don't mean anything, if you can have a dip in the curve just because JSL throws a hissy fit, or a peak because someone decides to do a "Get Ready Tune Competition"... :)
Oh and what Oswald said! |
| |
Mr. Mouse
Registered: Dec 2001
Posts: 235 |
Quote: Statistics like this don't mean anything, if you can have a dip in the curve just because JSL throws a hissy fit, or a peak because someone decides to do a "Get Ready Tune Competition"... :)
Oh and what Oswald said!
Statistics like this mean absolutely something, because you can actually now proceed to show that curves are due to events like you mention. That's what this is all about: pointing to the confounding factors and/or covariates that gave rise to the baseline curve. After all, the results above are "as is", raw data, and purely descriptive. Next up is looking into depth. :) |
| |
Frantic
Registered: Mar 2003
Posts: 1627 |
nice! |
| |
Mr. Mouse
Registered: Dec 2001
Posts: 235 |
Quote: Quoting Mr. MouseBack to topic, this is the distribution of the release types by release year, and it tells us the proportion of all the different types per year. Looks like especially large (relative) spikes were seen in graphics and music (and cracks) the past few years.
Nice! Could you do a distribution graph with absolute values as well?
I could, but perhaps this limited table may also be of some more use? It shows 20 Release types.
The 20% in- or decrease doesn't mean much with small numbers, but should have some impact on the overall number of releases when the Release type is one of these: Music, Graphics, One-file demo or Crack (those that have years with more than 100 releases).
The +400 overall increase in 2012 and 2013 each year can then largely be explained by an substantial increase in Cracks, Music, Graphics and One File Demos in 2012, with a further dramatic increase in Graphics and Music in 2013. The dip observed last year is then due to an equally dramatic decrease in Graphics and Music, not recovered by Cracks (that already went down substantially in 2013).
Taken together, for each of these "top" 4 Release types last year saw still a big number versus 2010. :) |
| |
Mr.Ammo
Account closed
Registered: Oct 2002
Posts: 228 |
Quoting Mr. Mouse
Anyway, what is somewhat troubling though, is that 2014 seems to be a year of decline.
Is that a gut feeling based on little over 1 quarter of data in 2015, or a hypothesis with some statistical significance? |
| |
TheRyk
Registered: Mar 2009
Posts: 2070 |
as I said before and will say again, 2013 was a peak of so-so significance.
nothing else can be oracled based on that stats so far,
no worries or as groepaz said nothing to see. |
| |
Mr. Mouse
Registered: Dec 2001
Posts: 235 |
Quote: Quoting Mr. Mouse
Anyway, what is somewhat troubling though, is that 2014 seems to be a year of decline.
Is that a gut feeling based on little over 1 quarter of data in 2015, or a hypothesis with some statistical significance?
Take heed of the word "seems". Also, read the other data. You will see that even though there is a decline versus 2013, it is still a big leap up from 2010. So no, I am not posing it as a decline that will continue downward as truth. I merely found it "somewhat troubling" and it "seemed" to be a year of decline. And technically, it actually is, versus previous year. Overall, we will have to wait for 2015 data to see the next point in the chart.
Nevertheless, the very fact that it can be pointed out makes it interesting to check the past 5 years. And that's what I am doing.
The biggest finding though is that 2012, 2013 and 2014 suggest an annual number of releases at a level we haven't seen since 1998 (as listed at CSDb). Personally, I think that is great news, and I wish this trend to continue! |
| |
Mr. Mouse
Registered: Dec 2001
Posts: 235 |
About those 2013 releases, zooming in on C64 Graphics, we can see part of the big increase was probably caused by the many graphics compos, such as Font Creation Competion 2013, C64pixels.com Double Screen Compo 2013, Plain PETSCII Graphics Competition and the 3-Color-Logo Competition 2013.
Here's a table with the top 15 "ReleasedBy" IDs as listed at CSDb (ranked for 2013, with the other years also in the table). As you can see, JSL was going strong both in 2013 and 2014 ;-)
But you can also see that a lot of those in the top 15 of 2013 did hardly release anything before or after that year, hence the correlation with the compos in 2013.
 |
| |
Mr. Mouse
Registered: Dec 2001
Posts: 235 |
Back to general, looking at the foundation years of the groups listed here, this chart comes up. You will see the number of unique new groups formed per year of foundation. The Golden Age C64 Year 1988 again sticks out! From about 2000 onwards the number of new groups each year seems to average out at about 14.5.
1988. If anything, a competition honoring 1988 would be fitting. Why don't we call 2016 the year of 1988? 1988 needs a tribute, it is in all probability the year where c64 scene activity reached it's supremacy. |
| |
Krill
Registered: Apr 2002
Posts: 2840 |
No. |
| |
Mr. Mouse
Registered: Dec 2001
Posts: 235 |
1988 is the year. Anyway, will post some more details of the past years soon. |
| |
Mr.Ammo
Account closed
Registered: Oct 2002
Posts: 228 |
Looking at the table of post 21, I'd say JSL is quite an outlier. Have you considered leaving out this outlier in your data set? |
| |
Mr. Mouse
Registered: Dec 2001
Posts: 235 |
Cumulative increase in releases per month for 2010 through 2015 (august 14, 2015). This year seems to be on par so far with 2012 and 2014. |
| |
Tao
Registered: Aug 2002
Posts: 115 |
Quoting Mr. MouseAbout those 2013 releases, zooming in on C64 Graphics, we can see part of the big increase was probably caused by the many graphics compos, such as Font Creation Competion 2013, C64pixels.com Double Screen Compo 2013, Plain PETSCII Graphics Competition and the 3-Color-Logo Competition 2013.
Here's a table with the top 15 "ReleasedBy" IDs as listed at CSDb (ranked for 2013, with the other years also in the table). As you can see, JSL was going strong both in 2013 and 2014 ;-)
But you can also see that a lot of those in the top 15 of 2013 did hardly release anything before or after that year, hence the correlation with the compos in 2013.
I didn't double-check the others, but most of the counts for TRIAD seem wrong: I checked out release page, and got the following counts for Graphics for those years:
2010 - 1
2011 - 3
2012 - 1
2013 - 12
2014 - 4
Some of our releases are credited to both artist & group; could this be the cause of the discrepancy? |
| |
Mr. Mouse
Registered: Dec 2001
Posts: 235 |
It could also be that some time later than the date this table was created new additions to CSDb were done. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11114 |
JSL wins!
what else do you need to know |
| |
TheRyk
Registered: Mar 2009
Posts: 2070 |
Alternative 1337 facts: Any stats mentioning MYD is fake news! ;) |
| |
hedning
Registered: Mar 2009
Posts: 4595 |
Mr. Mouse made an update today. It's really cool:
http://c64.xentax.com/media/csdb_quantifyme_2018.pdf |
| |
Bob
Registered: Nov 2002
Posts: 71 |
This is awesome!
Waaaah 2018 was a great scene year! \o/
And we won't get younger ;) |
| |
Raistlin
Registered: Mar 2007
Posts: 557 |
When Excess is mentioned in the charts at the end in the charts for, for example, 1982-1997.. is that the original Excess? Or New Excess? Or have both been combined? |
| |
Dano
Registered: Jul 2004
Posts: 226 |
I for myself would be interested who is currently the oldest and youngest scener around.
A demoscene release in 2018 from someone born before 1950 is pretty epic to me. |
| |
hedning
Registered: Mar 2009
Posts: 4595 |
Quote: When Excess is mentioned in the charts at the end in the charts for, for example, 1982-1997.. is that the original Excess? Or New Excess? Or have both been combined?
They have different id's, so it should be impossible for them to be combined. Looking at the amount of releases, I'd say it's the German Excess in the charts. But you already knew that. ;) |
| |
Mr. Mouse
Registered: Dec 2001
Posts: 235 |
Quote: They have different id's, so it should be impossible for them to be combined. Looking at the amount of releases, I'd say it's the German Excess in the charts. But you already knew that. ;)
Indeed. Different id's, so no combination. |
| |
Burglar
Registered: Dec 2004
Posts: 1031 |
cool stuff mr mouse!
I got 1 feature request: average csdb vote score per year, so we could see trends in release-quality too. (maybe only count releases with more than say 10 votes, to eliminate circle-jerk voting a bit?) |
| |
chatGPZ
Registered: Dec 2001
Posts: 11114 |
and a chart of who votes 10 or 1 most! |
| |
Oswald
Registered: Apr 2002
Posts: 5017 |
Quote: cool stuff mr mouse!
I got 1 feature request: average csdb vote score per year, so we could see trends in release-quality too. (maybe only count releases with more than say 10 votes, to eliminate circle-jerk voting a bit?)
its cicrle jerk anyway, imho similar standard demos get much higher votes by each year. |
| |
Carrion
Registered: Feb 2009
Posts: 317 |
Quote: its cicrle jerk anyway, imho similar standard demos get much higher votes by each year.
really?
I'd though it's opposite. The top demos of, say 1990, are rather mediocre now but back then they were 10/10.
Some portal introduced scale change for voting. Pixeljoint comes to mind in my area of interest. It's like they added 11 to 10 possible points (they actually added 6 to 5 but anyway). Which allows to rescale all the votes after few years of quality improvement.
anyway...
gret material Mr. Mouse. Thanks!
And if you allow me to share my private thoughts I'd say:
Ten years ago I get back to demoscening. and my observation is that more and more people are coming back, there's more production each year and quality is unbelivably increasing X after X. |
| |
Mr.Ammo
Account closed
Registered: Oct 2002
Posts: 228 |
Quoting Burglarcool stuff mr mouse!
I got 1 feature request: average csdb vote score per year, so we could see trends in release-quality too. (maybe only count releases with more than say 10 votes, to eliminate circle-jerk voting a bit?)
Removing all reciprocal votes would be a good idea too. |
| |
Mr. Mouse
Registered: Dec 2001
Posts: 235 |
Alas, I don't have individual vote statistics per scener. You know that would lead to scene drama anyway. ;-)
I can however look at some aggregated vote statistics per release, Burglar. Will do that later this week and post it here. :)
Although I personally do not think this voting system is fair, nor transparent. So the analysis will not tell anything, because the algorithm behind it is unknown. So I basically wouldn't know what I was looking at.
That is why I did the top 10 table with the number of comments per release, at least you can see that way how much "engagement" there was with those releases.
@Mr Ammo: To remove reciprocal votes (and I assume you mean people voting to help each other constantly) may be a little tricky to do, since we would need to look at voting-behaviour and identify common patterns. And then what to do? Delete those? Call out the voters for their conduct?
I guess the only way forward would be to remove hidden voting and make it all public. |
| |
Mr. Mouse
Registered: Dec 2001
Posts: 235 |
There really haven't been much votes for releases in the heyday period, and of course that is to be expected since people tend to vote for new releases each day, and not those from decades back. As of 2006 there is a sharp increase in voting, but 2013 is extraordinary. And not in line with the trend. Like a vote boom in that year. Overall the average release rating does crawl up the last years, but there is also overlapping standard deviations. So that probably doesn't really mean a notable difference. |
| |
Mr.Ammo
Account closed
Registered: Oct 2002
Posts: 228 |
Quoting Mr. Mouse@Mr Ammo: To remove reciprocal votes (and I assume you mean people voting to help each other constantly) may be a little tricky to do, since we would need to look at voting-behaviour and identify common patterns. And then what to do? Delete those? Call out the voters for their conduct?
I guess the only way forward would be to remove hidden voting and make it all public.
By removing all public (non-anonymous) votes from your data set, you're likely to remove a significant amount (if not all) of reciprocal and socially desirable/acceptable votes. |
| |
chatGPZ
Registered: Dec 2001
Posts: 11114 |
lol, i really doubt it. in the first place you'll remove a lot of votes =D
but comparing public vs non public would be interesting anyway, if only to show it really doesnt matter (for any entry with a reasonable number of votes) |
| |
Mr. Mouse
Registered: Dec 2001
Posts: 235 |
Hmm, I tend to agree with Groepaz. Yes, by removing public votes you may eliminate votes that are done for show, but will also delete votes done by people that are simply honest and not privacy prone. I think the biggest bias may come from the hidden votes, by down/up voters that don't wish to be discovered as down/up voter, or as a friend of the releaser in question. |
