morpheus problem
 

O.K. So I have read plenty of messages that should convince anyone that this morpheus thing is a real problem. I have had all the same frustrations shared earlier. Why havent I read anything about a pending solution? I have been thrilled with Limewire but I am now afraid that it will be useless soon. Why havent we heard from Limewires creators about a solution??

On its way
 

We are working on grouping LimeWire Ultrapeers more tightly to solve a few problems. We have found that 94% of traffic on a LimeWire Ultrapeer is non-LimeWire traffic. There are a lot of nasty clients <and perhaps others> out there effectively spamming the network.

I have a few more longer term solutions in mind but tight grouping should at least improve the local network in the short term.

Thanks
-greg

Thank you very much for this reply.
I really hope that your grouping plan is implemented soon and works well.
I am interested in your comment about so much of the traffic being non-limewire. Isnt this the best reason to develop "filters" or some other software device to frustrate this other traffic? If the solution was issued as a upgrade to "pro", I think it is safe to say that those of us using the software (and the network) daily, would line up tp pay for the fix now that it is apparant that the whole thing is in jepordy.

Fscott

Another thought..
I have a cble connection ans several Mac's at home. I could wipe anything important from one of them and leave it on 24/7.
If you could verify enough individuals willing, like me, you perhaps could build a backbone of people to establish filters on.
I admit, I really dont know what Im talking about here, but my point is (as was mentioned elswhere in the forum) this is a community and there may be others out here willing to be part of the solution....
Fscott

I think that what gbildson said should NOT be done, that will lead to the segmentation of gnutella, and can break the network. None peer should be treated differently. Also Morpheus is not that bad, most of it's problems is that it wont count well the number of hosts, by reducing the amount of pongs. But it is based on Gnucleus, and Gnucleus is an excelent client that takes very little bandwith, and does a great job as ultrapeer. I think the best ultrapeers are Limewire and Gnucleus (thus also includes Morpheus).
Also take it positively, Morpheus brings Gnutella a greater userbase, so you will be able to find more files.
By the way I don't use any of those. I use Shareaza, which is great.

Re: On its way
 

At some point Gnutella will definitely need a way to block spammers/spoofers.

This should be built into the protocol - ignoring certain IP addresses for a few hours and propagation of those banned addresses among ultrapeers.

Without this, Gnutella will die within six months as all the results returned will be useless - either porn spam or RIAA fakes.

I have been noticing that when the "Total Files" (in the Statistics Window) goes down, I look at my connections and over half of them will be Morpheus. I keep 10 connections up. But my search results go way down. Now, as it stands, we have to enter each and every IP address to block them, right? Is there, or will there be, a way to block Vendors instead of just IP addresses?
Thanks.

It goes down, because Morpheus/Gnucleus don't waste your bandwith, just to tell you how many hosts are there.

Thanks :) but, I'm not concerned with how many hosts (total) are there. My problem is that the number of files/results go way down when there are alot of Morpheus hosts.

No one would even bother to look at the hosts total if there were not a problem with search hits. This is a fact unchanged by your admiration for Morpheus as a Gnutella client.

However, I have noticed that Morpheus ultrapeers need to "cook" a lot longer than Limewire. If I leave Limewire on all night, then in the morning there are usually nothing but Morpheus ultrapeers left. If I do a search, I get the usual number of hits.

But who wants to wait eight hours or so to get decent search results?

Hi,

Morpheus drops LimeWire hostcounts, so the network horizon statistic you are observing is wrong. It is especially wrong if you are connected to all morpheus ultrapeers.

Avi

So many morpheus UP is not really a problem in the program, there are that many because morpheus holds the largest user base. The problem is that some of their childs don't share that many files, maybe because they are on slow connections.
But the problem is not the Ultrapeers, I saw that each one of those has at least 200 childs. You can see this by browsing the network (a gnucleus feature), if you want to see it, just open your browser and type http://{morpheus client ip}:6346, you will see that most of them have lots of childs, and are well connected to other ultrapeers.

I browsed through the morpheus ultrapeers via http and the average number of child nodes was about 50-60. That isn't great but it's good enough.

Again, the point is NOT the accuracy of the number of hosts indicated or how a Morpheus UP does or does not return that info. to its children/leaves.

Also, since all flavors of Gnutella client can connect to Morpheus UP's, then it doesn't matter in this regard whether a lot of Morpheus users don't share files. They will also be connecting to LW ultrapeers. You can deal with them, at least as far as your own sharing is concern, with your filters.

The point is how Morpheus behaves as an ultrapeer versus Limewire. It is as simple as this: if you have Limewire UP's, you get more hits. If you have a lot of Morpheus UP's you get less. And if you have Gnucleus UP's you get the least.

It DOES matter if Morpheus users don't share files. Each Ultrapeer can handle only so many leaf nodes. Whether you're a leaf node or an Ultrapeer, you still have a horizon within the net. There are a finite number of nodes within your horizon. The more of those nodes that aren't sharing, the less you're going to see.

Assume for the moment that most of the people using LimeWire understand the point of P2P (i.e. they share), and most of the people using Morpheus really don't get it. Before Morpheus switched to gnutella, you had mostly people who were sharing. You connect to a LimeWire Ultrapeer and most of the other leaf nodes attached to that UP are sharing, and most of the leaf nodes attached to the other UPs are sharing. Your horizon was still the same and you still had the same number of nodes within that horizon (possibly even less given that there were fewer people participating), but almost everyone you could see was sharing something. That translated into lots of successful searches.

Today you've got two million Morpheus users, most of whom, at least for the sake of this argument, are not sharing. Prior to the Morpheus invasion it appears there were an average of 50,000 to 75,000 clients online. Following the invasion the average number of clients shot up to about 250,000. Doing the math, that gives us about 175,000 Morpheus users, or 70% of the total clients. So, take what used to be a target-rich environment within your horizon and replace 65% (assuming at least some of the Morpheus clients are sharing) of the clients with leeches, and that automatically translates into 65% fewer search hits.

Just to be sure I've bludgeoned home the point, let's try an analogy. Suppose you're in an apple orchard with trees planted at regular intervals in all directions. You're securely tethered to a tree in the middle of the orchard. The tether is just long enough that you can reach seven trees away from the one to which you are secured. All the trees around you are healthy and full of fruit. Now, imagine the same scenario, except two-thirds of the trees around you have no fruit at all. I ask you, in which scenario are you most likely to not go hungry?

The problem isn't with stopping the leeches from downloading. The problem is that they're taking up space that used to be occupied by clients who shared.

I believe there is a very simple solution. Let's ask the developers to group LimeWire nodes together closely.

If LimeWire's leaf nodes were to connect to LimeWire ultrapeers only and four of six connections of LimeWire ultrapeers were to other LimeWire ultrapeers many of our problems were solved.

It'd really be a very simple task, - just adding three or four lines to the source code and it is done.

Anybody remember the days of v1.7? I could run 14 lines and 10000 hosts and get around 1000 results when i searched for most things i like (even with category filters). now when i search for things like digitized video i am lucky if i get 10 good results even close to what i asked for (not counting PORN)

Yes, ultimately it matters whether Morpheus users share files now that it has become a Gnutella client. This fact, however, is not directly connected to how Morpheus performs as an ultrapeer now that it's here. Non-sharing users are also connecting to LW ultrapeers and contribute to the degradation of the network generally. Yet even with this state of affairs, LW performs better as an ultrapeer than Morpheus.

BTW, if hardly anyone shared on the old Morpheus network, what on earth was the incentive for the people who did?

There wasn't much tangible incentive for people to share. There was endless discussion about it, some pro, some con, but they never did add a leech filter. If you were concerned about the leeches, the best you could do was a reverse lookup on clients downloading from you and manually delete the ones who weren't sharing (enough to satisfy your sensibilities). The problem was the software was quite good at automatically restarting an interrupted download, so unless you had lots of people trying to download from you, the leech would be right back downloading again.

As far as why anyone shared, I'm sure the reasons are quite varied. Some people just wanted to make the whole thing work and realized that required some decent file sources. Some had delusions of altruism. I suspect most simply realized it was fair that they share at least what they'd downloaded from others.

http://mitglied.lycos.de/mdsgeist/LimeWire.zip

LimeWire version with the added ability to block gnutella connections to certain vendors.

If you don't want to connect to Morpheus anymore, this will solve your problems, because it has an option to disallow certain vendors from creating to you by allowing you to specify strings that will be matched against any servents user agent.

Why block Morpheus, there are other ways to find a smarter solutions for this problems. The problem is in how Gnucleus decides to become an Ultrapeer, I'll add a message I posted in another forum:

So a better solution would be to help reduce the amount of Morpheus Ultrapeer than blocking them.

dague12

Hey, I got an idea. Let's make this a toggleable choice. I really don't want to wait for people to figure this out because working it out with morpheus will take a long time and a heck of a lot of troubleshooting. Let's just give users the choice to only connect to limewire clients if they want and let that be that. When morpheus and limewire finally come to terms (in a software sense), then the feature can be removed. This is just going to drive everyone nuts and ruin the system until someone finds a quick fix now.

Also remember that quick fixes don't have to be permanent.

Fighting fire with fire...
 

Am I the only person that sees that phrase as being completely ridiculous? After all, if one does fight fire with fire, you get a bigger fire...causing more problems.

Still, it seems to be the most popular suggestion on here..."block Morpheus clients" is the popular suggestion, and admittedly, the most convenient solution.

Gnutella is a network that was created on the idea of sharing without limitation...if LimeWire (which I think everyone will agree seems to be the most efficient and well programmed client for the network currently) starts blocking Morpheus, what's next? I get BearShare clients sometimes as well...block them? Anyway, you see where it's heading. I've got a slightly different idea...

Howzabout the developers for LimeWire and the developers for Morpheus try to come to some agreement? After all, they really are the two super-powers in the world of Gnutella.

I have a slightly different reason for suggesting this...I run primarily on a Macintosh platform. If I start looking for Macintosh files when there are only Morpheus (an up until now Windows exclusive program...another thing they could learn from the folks at LW) servers up, how many Macintosh files do you think there are going to be listed? ZERO! Goose egg! Because Windows users think that Macintosh users (this is a really broad generalization, I know, but when 99% of encounters support the theory, you begin to accept it as fact) are totally retarded and useless. But we make stuff pretty (Mac users are predominantly artists).

Now, if Morpheus opened itself up to the Mac platform...that might change things. However, I have a feeling that unless someone attepmts to persuade them to do so, they won't even know that there's this whole other platform out there being used by a good 297 million users that, at present, aren't earning them any money.

Of course, that's one solution, and an extreme one at that. Here's another...does anyone think that the people at Morpheus might be persuaded to change their policies for their clients? Say, to enable the freeloader policy? Again, Gnutella is intended for sharing, not taking. In that sense, freeloaders SHOULD be prevented from accessing other users' files.

Professional collaberation and courtesy...and perhaps it's a dumb idea.