File talk:Layout engine usage share.svg

Moving from http://en.wikipedia.org/wiki/Image_talk:Layout_engine_usage_share.png

Others

Just curious, what is "Others" in 1997. It takes up 10%. That is a pretty big share. AlistairMcMillan 19:40, 12 May 2005 (UTC)

No idea. It was based on Behind the numbers: Browser market share. --minghong 10:01, 13 May 2005 (UTC)

Netscape Classic and Tasman not dead

At least in Germany according to Webhits Netscape 4 has 2% and Tasman at least 0.8 % (MSIE 5.16 0.1% + MSIE 5.23 0.3% MSIE 5.17 0.4%) market share. Even the English Wikipedia counted 1.4% of NS4 in February 2004 and 1.8% of Tasman in October 2004. --Hhielscher 00:08, 27 May 2005 (UTC)

We're talking about worldwide trend only. --minghong 08:03, 4 Jun 2005 (UTC)

Go along if you happen to have better data. The statistical data is wrong anyway because browsers behind proxies (common at least in Spain and China) are not counted. But even Usage share of web browsers states:

OneStat.com           20050427 Netscape 1.4 %
Janco Associates Inc. 200504   Netscape 0.92%

--Hhielscher 23:57, 5 Jun 2005 (UTC)

Hhielscher, I just had a look at the data in the context of the page; the Janco data is only for this year while for OneStat, did you notice how Netscape was in decline from 04/2002 through 07/2003, then disappeared for all of 2004 before showing up again? I suspect what's going on there is not that Netscape Classic is making a comeback, but rather that its reflecting the new Netscape (which is based on Gecko; what powers Firefox).

Consider that for Netscape Navigator, the last classic version of Netscape (4.8) was released 08/2002; everything after that (6+) is based on Gecko; 7.2 was released 08/2004 (and 8.0 was released about a week ago). Also, while Minghong didn't want to include it in the main page for the image, please see the W3Schools Browser Stats page; while not quantitative for the internet as a whole, they do nicely show qualitative trends. Note that Netscape 4.x started in 2002 with over 4%, but dropped to 2% by the end of the year and further dropped to half a percent by the end of 2003. The numbers have only been downhill from there; its at 0.2% as of May 2005, so no big upturn. Netscape 7.x had about 4x that for the same month (on the W3schools site, anyway).

--Limulus 13:32, 6 Jun 2005 (UTC)

Yup, that's exactly what I think. Those should be Netscape 6/7/8, not the Netscape Navigator/Communicator. --minghong 05:10, 8 Jun 2005 (UTC)

Also, our graph doesn't show Tasman, right? So we didn't say it is dead as it was never included in the graph. --minghong 05:11, 8 Jun 2005 (UTC)

Technically, it *is* in the graph, under "Others" ;) --Limulus 08:59, 23 July 2005 (UTC)

Edit: Tasman (layout engine) is IE for Mac (I have heard of it, but it wasn't a huge % of market share :) In keeping with my previous comments about NC/Gecko, I guess Tasman speciated from Trident? --Limulus 09:06, 23 July 2005 (UTC)

data before 1996?

What about using the GVU's WWW User Survey's?--Hhielscher 01:44, 27 May 2005 (UTC)

We're already using it. But this is only one source. We should have multiple sources so as to average the biases (or noises) introduced in each source. ;-) --minghong 08:02, 4 Jun 2005 (UTC)

Colorblind Friendly Image?

When I redid the image, I didn't use the hatching that made it "more colorblind friendly"; personally I think it looks better this way, but if there are any objections, please leave a message here :)

I'm going to redraw with hatching. --minghong 08:02, 4 Jun 2005 (UTC)

Please then also update the SXC file. Thanks! --Limulus 09:25, 4 Jun 2005 (UTC)

I'd love to. But I don't know where you uploaded it. Wikicommon? --minghong 16:43, 4 Jun 2005 (UTC)

No; I linked it (Layout engine usage share.sxc) directly after the 'suggestions for improvements' section:

Graph created with OpenOffice.org Calc (Spreadsheet): Media:Layout engine usage share.sxc

Click the second link and you'll get the file I was working on; it has everything except for the cross-hatching. Of course, if you prefer, just upload your SXC file :) --Limulus 06:18, 5 Jun 2005 (UTC)

I thought Wikipedia won't allow me to upload non-image/audio files... Maybe they changed the policy? Anyway, I uploaded the updated sxc file. --minghong 05:08, 8 Jun 2005 (UTC)

Hmm... I just found a neat online tool; the Colorblind Web Page Filter which simulates what images look like to the colorblind; I ran my version of the image through it (with the various filters) and I think it still looks pretty good... --Limulus 22:11, 5 Jun 2005 (UTC)

January 1

In the image page it now said "Data sould be presented in a way such that the year marker indicates data for January 1 of that year (e.g. reading up from "2005" should indicate approximate market share for Jan. 1, 2005).". The problem of this is that many sources doesn't give the statistics regularly. We only got random reports. So I thought using the average values of those reports should be good enough. --minghong 08:33, 4 Jun 2005 (UTC)

The way I thought about it, if you have an 'average value' for a year, that's roughly the same as for the middle of the year, so if you position the two data points in such a way as to pass through that point it all works out (just my take on the matter :-) Hope you don't terribly mind the changes I've been making! --Limulus 09:19, 4 Jun 2005 (UTC)

NPOV?

I am not at all convinced that the presentation of this graph represents a neutral point of view. The graph is obviously meant to visualise a pattern that when a newer rendering engine is introduced, its market share rises until it gains total dominance and wipes out the previous market leader, and this cycle repeats, the implication presumably being that Firefox will be the latest to continue this trend. Particular problems I have are:

1. the graph could plot each browser's usage share as a seperate line, rather than having new browsers crush old browsers with their colourful gravity

2. the choice of how to seperate rendering engines is entirely arbitrary. Firefox is originally based on Netscape code so why is Netscape Classic entirely seperate in the graph from Firefox, whilst Internet Explorer is always counted as a single rendering engine?

3. Why does Opera not exist in the graph until 2003?

If these 3 changes were made, the graph would be entirely different - it would show that the relative share of IE and Netscape based browsers falls and rises over time, with Netscape more popular in the past and IE more popular in the present day, whilst smaller browsers (include Safari and Opera in the present day, and various small projects as well as Opera in the past) have always had and continue to have a small but steady presence. --62.181.235.221 21:41, 15 July 2005 (UTC)

Hi MS Peon.

2 - Firefox is not at all based on Netscape code. The hole point of creating Firefox was to start from the zero. I suggest you google to learn more.

I am against your claim. --Sekelsenmat 03:16, 17 July 2005 (UTC)

Hey, 3 more points from the MS Peon:

1. If you were to make a graph of the relative success of different Formula 1 teams over the years, would you seperate the graph on the basis of when the teams rebuilt the cars, or what the teams were called? If you were to make a graph of the relative success of different political parties at elections, would you start again each time the leader has been replaced, or once a certain proportion of the party's representatives had changed?

2. Firefox wasn't a direct fork of Netscape code, but what did happen was: a). Nescape paid people to make a browser b) They released the source code to their browser c) The OS community looked at the source d) a mixture of Netscape employees who had written the old browser and OS community people who had read the source to the old browser wrote a new one. Yes, it was rewritten, but to say that they "started from zero" is an exagerration.

3. It is only natural that a) People write in Wikipedia about things that interest them and b) Things that people are interested in and things that they have a strong opinion on are often closely corellated. However, the point of wikipedia is not to be a political soapbox. Even if it is true that the world would be a better place if people Spread Firefox, that doesn't matter. The people who manage this article are evidently Firefox fans (note I didn't say "peons" or "zealots" or some other derogatory term), but they should keep their campaign to Spread Firefox to the rest of the web, not a website whose express purpose is to be neutral.

4. You didn't even try to argue against my earlier points (1) and (3). --62.181.235.221 13:27, 20 July 2005 (UTC)

Um, ok, my $0.02...

1. The graph could be redone in lots of ways (I'm thinking that the easiest way that it could be redone the way you like is changing the graph type to 'Normal' and playing with Transparancies... But I'd have to poke at it for a bit) but seriously, Mosaic was wiped out by Netscape, which was then CRUSHED by Internet Explorer. That is historical fact as far as I am aware :)

Now while you say that the graph is biased to make it look like FF is going to crush IE, what does that say about Opera and Safari? That they are going to crush FF? Like those investment companies say, 'past performance is no guarantee of future results'...

2. Regarding the choice of rendering engines, I consider Netscape Classic and Gecko to have speciated; while Gecko is descended from NC (which is now pretty much extinct in the wild ;), Gecko went on its own path and changed significantly. The same cannot be said for Trident AFAIK.

3. As far as Opera share goes, remember first that the graph is by rendering engine, not program name; the RE listed for Opera, namely Presto, is only for Opera 7+; prior to that, Opera would be under "Others". Secondly, go back to the sources used for the image, namely Usage_share_of_web_browsers. You'll note that Opera doesn't show up with any significance until 2002 BUT Opera 7 wasn't released until 2003.

4. (3b ;) While the other people working on this image may be FF fans (I admit I am :), I know that I tried very hard to find objective numbers for my contribution. The simple fact of the matter is that Gecko (Firefox + Mozilla + Netscape 6+) has (currently) nibbled away at IE to the tune of 10+% with FF being the bulk of that; according to a recent NetApplications usage share study, FF is grabbing up to a percent monthly and its basically all coming from IE. Meanwhile, Safari and Opera are fairly flat. This is what the graph current graph indicates.

So basically, I reject all of your points except #1 (while I don't think its biased I can see your POV)

P.S. I really like your phrase "new browsers crush old browsers with their colourful gravity" (its apt and cute in a Katamari_Damacy kind of way ^_^)

--Limulus 08:50, 23 July 2005 (UTC)

OK, what I have done is change the graph part in File:Layout engine usage share.sxc But as I feared, its rather ugly and doesn't do near as good a job of presenting the data. Please observe that:

- It illustrates Netscape replacing Mosaic well.

- It shows the Browser Wars (IE v NN) well.

- It does a good job in showing IE's decline since its peak circa 2002.

- It does a mediocre job of showing Gecko stealing browser share from IE.

- It does an awful job at showing Safari/Opera/Others (since ~2003) because they're so small and so close in value.

The solution IMHO would be to either:

(1) use the new graphing technique, but split up the graph into two (one showing Mosaic being replaced by NN being replaced by IE and the second showing detail of minority browser trends since ~2001)

(2) use the new graphing technique, but remove Safari and Opera altogether (that is, incorporate their data into Others) since they're rather minor and not increasing, or

(3) keep the current graph since it does a good job of incorporating all the data visually ;)

I submit to 62.181.235.221 that the current trend is in fact a downward one for IE... BUT that if it reverses or stabilizes, with the de facto graphing technique it will display that nicely in future versions of the graph. Is that enough for neutrality of the image to be re-established?

--Limulus 10:32, 23 July 2005 (UTC)

I think the current graph is ok (well, after all, I'm the creator). The engines are sorted by their creation dates, so this is definately not a POV. And the new graph looks pretty awful: there are too much overlapping that hide the minority engines. In the current graph, we can still see several thin lines. --minghong 10:57, 23 July 2005 (UTC)

So do we give 62.181.235.221 some time to respond or should we just delete the POV tag from the image page? Oh and BTW, that reminds me that when we do that I have to fix the image (and SXC file when we revert it) because of the little mistake I made in the CSV data ^_-; Just a little note to myself... --Limulus 11:50, 23 July 2005 (UTC)

Just give everybody some time. We are not in hurry, right? --minghong 15:13, 25 July 2005 (UTC)

To 62.181.235.221:

2 - From http://en.wikipedia.org/wiki/History_of_Mozilla_Application_Suite#Open_sourcing_of_Communicator "Ultimately, the Mozilla core developers concluded that the old code could not be salvaged. As stated on the October 26, 1998 development roadmap, it was decided to scrap the whole code base and rewrite it from the ground up. This caused one of the former lead Netscape developers, Jamie Zawinski, to resign. The resulting plan included, among other things, the creation of a whole new cross-platform user interface library and a new layout engine."

Sorry, but Mozilla was rebuilt from zero. You will reach similar conclusions googling. Mozilla != Netscape.

4 - I did not write about your points 1 and 3 because I had nothing to add to them. As I expected others have already answered that.

I am in favor of removing the NPOV. --Sekelsenmat 18:39, 26 July 2005 (UTC)

I don't dispute that FF has seen a significant increase in share over the past few years, and obviously any graph should reflect that as the main story in the past few years. However, my argument is that the graph obviously implies that when released newer browsers obsolete older browsers. But this isn't the case because:

1. Only the optimistic Firefox fans really believe they are going to wipe out Microsoft 2. IE has been heavily updated on a number of occasions since its release, such that IE 6 is practically unrecognisable when compared to IE 1 (see http://en.wikipedia.org/wiki/Image:Megara_008.jpg for a screenshot of IE 1.5) 3. The first version of Opera was released about the same time as the first version of IE, and at no point have Opera simply thrown out all their old code in one go 4. Though FF may not use Netscape code (which I already said before Sekelsenmat proved me "wrong" be repeating exactly the same thing again), the Mozilla effort started out of Netscape

I just had a quick play with Open Office and I admit it is hard to make a graph with individual line plots for each browser that is not very messy. I may come back and have another go at some point. But still, the graph as it currently stands looks like it could better be titled "The rise of Firefox" than "Layout engine usage share", and I'd imagine changing the graph would be nicer than changing the title. 62.181.235.221 11:52, 27 July 2005 (UTC) --

This does not make sense. The graph is based on real data colection. The graph should not be changed just because someone thinks that it implies something about the future and that it should imply something else about the future. The graph should not be changed to fit into a prediction of the future that Firefox will not grow past a point. --Felipe Monteiro de Carvalho 16:58, 10 September 2005 (UTC)

Graphs, in my opinion, should not try to predict the future without evidence. This graph indeed provides some evidence that Firefox is replacing Internet Explorer as the chief browser, but the graph could be manipulated to imply that Firefox has stabilized and is now being replaced by Opera (I'll add an image to prove this point). I think it may be too soon in Firefox's rise, which I think there is significant evidence that it is still occurring, to have a graph that implies this rise; there simply may not be enough data. I'm sure there are objective ways of representing this data - and again, objective does not neccesarily mean that the graph would not try to show the pattern of the rise and fall of the different browsers. Theshibboleth 06:03, 29 October 2005 (UTC)

I was unfamiliar with OpenOffice's graphing capabilities, and that the graph as presented was rendered by a computer. It probably is the best way to convey the data, though the colors are somewhat awkward, and the crosshatching gives the image a somewhat unproffessional look, though I suppose the cross-hathing's good since it allows people without color monitors to still be able to read the graph. I might upload an updated version that is still color blind-friendly. Theshibboleth 06:16, 29 October 2005 (UTC)

Crosshatching or no crosshatching?

I've uploaded a new version of the image. I really hate the crosshatching, and I would recommend that we conduct a vote on the matter, though I'm not exactly sure how to do this. Theshibboleth 07:06, 29 October 2005 (UTC)

Data

How are the values used for this graph calculated from the various tables on the browser stats page? Amaurea 00:41, 8 January 2006 (UTC)

See the page and the talk page for "Layout_engine_usage_share.png"; as I recall, the (often somewhat conflicting) numbers (especially the early ones) from Usage_share_of_web_browsers were eyeballed and rounded versions were produced that showed the trends more smoothly. There is a certain amount of art in producing a graph like this since even today its hard to pin down very accurate numbers, though trends become obvious fairly quickly...

Consider the question: "What percentage does Firefox represent of browsers used?" On that page numbers for late 2005 range from ~8% to ~12.5% (the high numbers are 50% larger than the small ones!) So for a graph point we might just say 10%. Similarly in late 2004 the range was from ~3% to ~5.5% (the high numbers are nearly double the low ones!) so we might say 4%, but clearly the *trend* from late 2004 to late 2005 is that Firefox has greatly increased its share. This sort of qualitative trend is supported by sites like W3Schools which aren't indicative of the web as a whole, but do clearly indicate transitions.

Then of course, if you start looking at other sources, like XitiMonitor, you get more (conflicting) data to try to fit into your graph :) Hope that answers your question. -- Limulus 05:23, 8 January 2006 (UTC)

Convert to svg

This graph should be turned into svg, but i really can't se how the layout should be. Collaboration to convert graphs to SVG says that the graph should be in grayscale, for more useable prints and photocopies.

Should we keep the layout, or should we make a new one? Henrikb4 19:13, 16 August 2006 (UTC)

I've been planning on re-doing the graph in SVG. I see some disagreement about the use of grayscale, so how about using the existing colors without the crosshatching? -- Schapel 18:14, 21 August 2006 (UTC)

CVS Data seems incorrect for 2006

I was working on a svg version when I discovered that the values for 2006 sum to 102.2025. I would like to fix this, but I'm not very familiar with the data sources. Please will someone help me fix this. Gary van der Merwe 10:05, 31 August 2006 (UTC)

I'm planning on redoing all the data after the Q3 2006 results are in from thecounter.com and Net Applications. -- Schapel 11:11, 31 August 2006 (UTC)

Moving completed

In addition, it is practically inconceivable that Safari's share would have jumped from under 1.6% in 2005 to almost 31% in 2006. —Angr 16:13, 21 April 2007 (UTC)

Chart junk/vibration

There is a lot of unnecessary chartjunk in this graph: the black lines between areas and the lines inside the areas. --Qubodup (talk) 11:09, 14 February 2009 (UTC)