Skype Hits 40 Million Simultaneous Users | Hacker News Reader

Skype Hits 40 Million Simultaneous Users

disruptivetelephony.com

59 pointsdanyork43 comments

I'm surprised to see the complaints about call quality. In recent years I've found the call quality to be really good. It helps if you have a dedicated headset.

I'd much rather criticize their UI design. The latest OSX client is an abomination.

I am also surprised about the complaints. I use Skype for calls multiple times each week using both voice and video and very seldom have any quality issues. Every now and then I have glitches, but for the most part the quality is great.

And yes, I agree with you about the 5.x OS X client (and have written multiple articles about the deficiencies it has).

If people need something to criticize, I'd much rather criticize Skype's lack of IPv6 support, which is going to be problematic as more networks (including my own) look at moving to using more IPv6.

I switched from Windows to OS X a couple months ago. Everything was bright and shiny until I installed the Skype client. I could literally not believe that this is the same Software I used to love on Windows - there is so much wrong with it (constantly popping up Dial Pad that I never use, inability to resize the Name column, to just name a few). Over time I got used to the annoyances, but it it still one of the worst UIs I've ever seen. Does anybody have an idea how this could happen?

The "craigslist effect". It's when a product or service with shitty design and severe flaws is the only one that does what it does and starts growing in popularity due to it's monopoly and it's owners think "hey we must be doing something right so lets not change or improve it because we don't have to". A.K.A. "What happens when you don't have competition."

But what's so surprising is that I don't remember the Windows client to be nearly as bad as the one for Mac. I guess they are made by two completely different teams with no coordination on UI issues.

The Mac Skype client was actually quite excellent, too, up through version 2.8. Then Skype jumped the version number to 5.0 and came out with the UI mess that they did. About a year ago I wrote about the issues I had with the 5.x client... and I still have the issue with the lack of multiple windows:

http://www.disruptivetelephony.com/2011/04/my-3-main-issues-...

And yes, Skype's clients have historically been made by completely different product teams with the result being a very fragmented product strategy. Different features have been released at different times on different platforms, with the result being that users have very different experiences from platform to platform.

They are trying to get better at this and with their latest 5.x releases are trying to come out on at least Windows and Mac OS X at the same time with similar features.

I downgraded (and therefore "upgraded") back to Skype 2.8.0.866 on the Mac a long time ago. That version actually has a sensible UI compared to Skype 5.

If you search for "Skype 2.8.0.866" you'll find sites that host the original version.

Unfortunately, screen sharing no longer works in Skype 2.8 on Lion.

I have issues with call quality, and many others do too, but it's not a fault with Skype.

I'm just on unreliable and quite slow DSL with a constrained upload/download ratio.

Maybe they could work on the quality of the service now!

One of my largest pet peeves is that Skype quality is so poor that DTMF tones can't travel across intact. If you're using Skype and trying to enter an extension, good luck. This morning I tried to use Skype to access my voice mail without success.

Customer support is also horrible. The standard response is that it is the customer's fault. Audio quality on Skype seems to be directly proportional to how old the device is.

Nokia Maemo tablet? Great! Nokia N95, not so great. Windows, Android, Linux, Apple? Nuh uh.

> One of my largest pet peeves is that Skype quality is so poor that DTMF tones can't travel across intact

I nearly threw my laptop across the room in frustration this morning due to this issue.

I was trying to log in to a conference call. I think it took me 5 calls and 9 attempts.

Does anyone know the technical reason for DTMF tones having difficulty?

Because the quality of a Skype call, for me, is so much higher than a cell phone or landline, why do the tones have a problem? Is it something about the codec?

(https://support.skype.com/en-us/faq/FA10292/Does-Skype-Conne...)

(http://datatracker.ietf.org/doc/rfc2833/)

(http://datatracker.ietf.org/doc/rfc4733/)

(http://datatracker.ietf.org/doc/rfc4734/)

(https://lists.cs.columbia.edu/pipermail/sip-implementors/200...)

> GK>> You are right that with in-band DTMF, you need to perform tone detection at RTP level which isn't exactly an ideal scenario.

(http://www.voipmechanic.com/dtmf-issues.htm)

> In some VoIP routes a switch may be configured to detect in-band DTMF which is sent by the VoIP ATA, but then switches to an out of band RFC2833 DTMF required for an upstream provider. This upstream carrier then terminates the call to the PSTN, possibly to a voice mail system, which will require regeneration of the audible inband DTMF tones. The switch has to detect and remove the tone sent by the ATA from the audio stream because the upstream provider specified RFC2833 DTMF. At times the switch can't always completely remove the in-band DTMF tone which is a problem, because by the time it has detected the DTMF tone, it has already passed a short amount of it. This small amount of in-band tone along with the RFC2833 tone sent are both received by the far end voice mail system which will then register an error (problem), possibly an invalid mailbox or invalid password.

Ahhh, that would explain the problems I had recently trying to join a conference call. It took me 4-5 times, trying varying amounts of time between presses, before the conference system let me in. Voice wise, I rarely have a problem - even doing Australia to east coast USA - apart from the slight delay in travelling half way around the world.

I never seem to have these quality issues that people are complaining about elsewhere in this thread. About the only thing I can give them flack for is the horrible, horrible UI design.

Are you talking about the windows or OSX one? They're both so different.

Both of them are pretty painful, albeit in different ways. OSX Skype could take a cue from Adium, Windows Skype could take a cue from ICQ, and BOTH could take a cue from Linux Skype.

Linux Skype, which makes you open a new window for every chat you have? Linux Skype which won't notify you of incoming calls if you're already on a call? That Linux Skype?

I was talking more about the minimalist UI and less about janky behavior. The Windows (and to a lesser extent, OS X) versions seem to have taken their designs from the worst parts of MSN Messenger and AIM.

And call me crazy, but I prefer my chats in separate windows.

While I congratulate Skype on reaching 40M simultaneous users, I am somewhat disappointed that there was no mention of how the 40M people can stay online without melting Skype's servers. Some architectural insight would have been nice from an article posted on the front page of HN...

Skype uses a peer-to-peer network for the majority of the connections and really only uses servers for authentication and for gateways to other services/networks (such as the regular phone network).

A couple of years ago I wrote a brief description of the technology underlying Skype:

http://www.disruptivetelephony.com/2010/11/a-brief-primer-on...

(I have no connection to Skype myself, but have worked in the VoIP industry for 10 years and have been writing about Skype for a good portion of that time.)

Cool, thanks for the link. If I understood correctly from your article, supernodes are basically Skype clients that act as distributed servers used for discovery of other clients.

But I imagine authentication has to happen against a Skype server, and the clients would have to constantly communicate with that server to stay online, to get the current balance, etc... That's 40M requests at any given time. Since it's semi-distributed, and you have to pay for the service (to call a landline or internationally, for example), how does this fit into the distributed model and how do they keep track of everyone?

Skype only gives the appearance of being centralised. It's almost all P2P.

Skype uses P2P to handle the majority of "call" data, so the load on their servers increases less than proportionately as the number of users increases.

Which means there were probably 100 million users attempting to use Skype simultaneously. Skype works maybe 50% of the time I try to use it at best.

Maybe I'll finally have to see if I can do something interesting with skype.

http://search.cpan.org/~mncoppola/Win32-Skype-0.01/lib/Win32...

I wrote a script that lets you initiate and respond to calls using only the keyboard, which made my Skype usage bearable when I still had to use it: https://raw.github.com/technomancy/dotfiles/master/bin/skyyy

There's Skypekit, an official API distributed by Skype themselves, with the unfortunate restriction that it cannot be used in any OSS apps.

Incidentally, this is why I'm using Trillian and not Adium right now...

Just to clear things up, that's 40M people logged into a Skype client, not simultaneously video chatting.

That sounds really high? That's more than 10% of the population of the USA, all using Skype at once.

Skype is used worldwide. So when you look at it as 0.589% of the world population, it seems a bit more realistic.

That's the number of people who have skype running on their computers, not the number who are on a call. Skype is fairly common, automatically starts when the computer starts, and doesn't actually close when you click the "x". So I think that number is believable for total users online worldwide.

Not to mention smartphones.

Are there rooms for an alternative?

Of course there is. I don't think it will happen, however. Google Plus already demonstrates rich in-browser video teleconferencing. I think that's the way to go.

It should be relatively easy to build a Google hangout client with nothing but HTML and a WebKit+plugins engine.

Could you be more specific on why this will not happen? As you say, in-web calls should be possible, why people don't compete against skype with decent web apps?

I think a new stand-alone application has zero chance of successfully taking on Skype head to head. A web app, possibly embedded in a social setting, has a much better chance of success, if well executed.

As far as I'm concerned, (Linux) Skype should die. It's the worst thing running on my PC right now, and I really hate running it (it's not set to autostart). It will usually hang insidiously every few minutes, and it's not even obvious. Everything continues to work, but messages won't get delivered, calls won't get answered and there's no notification of incoming calls.

The only way for me to notice that my Skype has frozen is to notice the 100% CPU usage on the task manager. I would love something better coming along, I look forward to the day when I never have to use Skype again.

I think Microsoft got a good deal when they bought skype.

Looking at Skype (despite it's flaws) it's success was unavoidable. The same way people want an all-in-one device that does everything (like smart phones) over multiple separate solutions (camera, mp3 player, phone, pda), people seem to want one IM client for calling, texting, IMing, voice chat & video chatting. I don't think any other client provides everything wrapped up so well. Yes the UI is horrendous and tedious but its services are well offered and cheap. And with ever increasing globalization those services are vital.

It's just too bad that Microsoft introduced a back door to spy on skype calls. I now trust it about as much as I trust Google Voice + Google Talk.

Please back the backdoor claim up, it's the first time I've read that and it does not really make sense (to me), as there would be no reason to do so...

http://www.computerworld.com/s/article/9218002/Microsoft_see...

Correction: Not a back door, but a eavesdropping technology.

What? What's the backdoor?

Reply on news.ycombinator.com