When companies try to hide information about their products and services from consumers, it’s a black and white case: let’s hammer them to be more transparent.
Unfortunately, recent calls for Singapore’s telecom operators to declare an “average” surfing speed or “minimum guaranteed speed” are as misguided as they are unhelpful in solving the biggest problems users face here – a lack of content and inconsistent links to overseas sites.
Trust me when I say I’m just as frustrated a broadband user as the next here, but many of these calls come from a lack of understanding about how the Internet works, and what really plagues the average user at home. Rather than get to the root of the problem, they obfuscate the situation and muddle things.
First of all, the government regulator here already publishes real upload and download speeds, known as throughput, on its website, so folks asking for more transparency should really first have a look at that and see what’s missing and what should be included.
And to explain broadband speeds, it helps to look at another Singapore obsession: cars.
Even if you buy an Audi R8 supercar capable of 300kmh or 400kmh, you still travel on the same roads here as the average Kia or Toyota driver, which means you suffer from the same jams on the expressways here during peak-hour traffic. Result: 30km/h or less on busy roads.
Does that mean that Audi should thus sell its cars with a “minimum guaranteed speed” of 30km/h on its brochures instead of the 300km/h that the engine is capable of? Would it be deceiving drivers who readily fork out several times the cost of a Toyota or Kia?
If that seems absurd, then the current calls for so-called “real” broadband speeds like “average surfing speed” or “minimum guaranteed speed” in telcos’ advertisements do not make sense either.
Just like Audi cannot guarantee how fast you can drive your car over different roads, your telco cannot guarantee a promised average or minimum speed to every single site its users surf to because they do not own every single site on the Net and cannot buy a speedy link to all sites.
The main reason why many sites, say, a Tudou.com or Download.com, is slow is because many of these overseas sites do not deliver their content at such high speeds that we enjoy over our broadband connection.
Even if you have a 100Mbps fibre optic broadband service, it won’t do any good if the max that the website you are downloading the file from offers just 300Kbps to each user, depending on the load that the server allows. What can your Singapore broadband operator do about that? Not much, unless it gets the content or website provider to locate its data in Singapore.
Another point that has been raised is that telcos are not being upfront about download speeds. But really, are they not? The most important content – anti-virus updates, driver updates and even Windows updates – are now available at much faster speeds than a few years ago, and often close to the speeds promised on paper.
Don’t believe that? Try downloading a game from the EA Singapore website, or a site that has “mirrored” or “cached” its content here, such as AMD and Nvidia, whose drivers are downloaded all the time by users here.
I subscribe to a 16Mbps StarHub link, which gives me as much as 1.4MB per second for such links. That’s roughly 11.2Mbps – close enough to the 16Mbps that StarHub promises, considering I still have several other apps running in the background.
There are many things I don’t like about my StarHub connection – like slow links during peak hours, but more on that later – yet, I won’t be so fast to say StarHub or indeed any service provider in Singapore is not “coming clean” or “deceiving” customers.
That’s because they do deliver near the top speed when the content is available to the speedy network here. The problem, as always with a small market like Singapore, has been local content, and while it is trickling in, there are still thousands of websites that are slow to our users here.
Blaming SingTel and StarHub is just an easy way to find a scapegoat without looking at the bigger problem, which is getting more content providers to base or cache their contents here via content distributors such as Akamai.
This is not to say the IDA (Infocomm Development Authority) should be lenient towards the telcos, or that nothing should be done. On the contrary, the IDA should be more stringent, but in asking for transparency, it should ask for the right details, not a halfway-house solution.
Let’s stop for a moment and ask what is this transparency that we are demanding for. Should telcos advertise an “average” speed based on where its users surf to each month, considering that this changes because users surf to different places offering varying download speeds each month?
Does this not confuse users looking for a guide to how fast a service is? Think about the car analogy again: imagine going to the Audi showroom and finding that this month, the car is capable of only an average 30km/h, while last week it was 50km/h. The hardware is still the same, and it can really run at a top speed of 300km/h, but hey, people want “real” speeds.
If that sounds absurd, then you’ve understood where this recent speed debate is going. It’s barking up the wrong tree.
Top speeds should remain as a guide to what is on offer. And if the IDA, which first threw up this question last month by asking if it should make telcos reveal “real” speeds, really wants to improve broadband speeds here, it should look at things in detail, and not to grab some populist headlines in the newspapers, at the expense of the telcos it regulates.
Here’s how.
Instead of a meaningless figure, the IDA should get telcos to show how fast they link up to, say, the top 10 sites that users here go to. If these throughput tests are run by IDA, they should give a transparent, unbiased view of which telco is giving users the best bang for buck.
Let the market do its work here. When telcos find that users are being slowed down when they access, say, Tudou.com or their World of Warcraft server, they will have to either expand their overseas bandwidth or somehow strike a deal with the content providers to base their content here, to keep users happy.
Make it mandatory for every telco to publish this chart of how fast it performs against a competitor in its advertisements, and you will get many Internet engineers at SingTel and StarHub pulling their socks up in no time.
The whole point is in making sure the consumer gets the “real speed” that a service provider offers. It’s not about a confusing “average” or “minimum guaranteed speed”, but a snapshot of what a telco actually offers every month to the top online destinations that users go to.
Just like on the roads, no one can guarantee you a promised speed on the Net. But you can make choices based on experience. If you are serious in studying how fast a broadband service really is, just follow its published speeds over a few months and you’ll get the information to make a better buying decision.
@Chad
You bring up an interesting point that I agree and could be the source of discontent with these published “up to speed”. I’ve personally experienced arbitrary speed capping that also increased latency in the process.
This is totally non-transparent and it is also not documented in any fine print contract with Starhub. Complains about these are usually attributed to “line noise” that would disappear mysteriously.
Apart from encouraging “premium” prices from ISPs, I think IDA could spend some time auditing them on these arbitrary capping. Having them come clean about these speed caps after exceeding download limits would certainly make all these debates more productive. They should also come clean and clearly indicate that they do (or do not) cap P2P traffic.
Transparency to funny business that do they on the back end, and options for premium dedicated plans should be the focus of IDA.
Alfred – yeah, I agree with all your points here. However I would distinguish between shaping/peak-time management(which is network wide and affects everyone using a service) and selective individual user rate limiting. The former I deem as normal network management (although more transparency would be good here too), the latter is an individual user T+C policy and if a user pays for 100mbit peak speed, consumer laws (in most Western countries, at least) would dictate I must be very clear if I am going to deny that service and force it down to 10mbit for an individual customer due to a policy.
There is no such public policy here that I can find. There are no data caps, but if telcos are going to apply them on individuals after a pre-determined point they surely have an obligation to make it crystal clear to the users at which point it applied.
This issue is magnified when you consider the tendency of carriers here to lock people into long-term contracts (1 or 2 years). Consider it from the perspective of someone buying a 100mbit plan – obviously I wouldn’t pay for this if I didn’t have a requirement for reasonably large amounts of data. But a few months into my 2 year contract, the telco changes the rules (without telling anyone) and starts rate limiting at an unspecified data cap. This is commercially and ethically wrong. This is kinda like selling someone a car with an extended 5 year warranty, and after a year telling customers the warranty only applies if your drive less than 5,000km a year.
Now I’m stuck paying for a different service to that I thought I was getting, and I have no comeback? I’m stuck for the rest of my contract. The silly thing is that I’d pay more than what I currently am – but the confusion and lack of certainty around what one is getting incentivises me to pay less to avoid feeling ripped off or locked in at a high rate with diminished service. 🙁
Thanks for pointing that out, Chad. Yes, there’s a difference in the areas you mentioned. Agree with you there: these are the areas that I feel the regulator has to come in hard on telcos. 🙂
I posted this comment because I relate to it because of my work. I do not mean any offense to anyone.
Throughput is defined as the average rate of transmission over a network. When you say the throughput of a network is 2Mbps, it means that the average transmission rate is 2Mbps. So essentially, both Alfred and CK is asking for the same thing. It is important to note that, as CK rightly pointed out, throughput is an average value. Actual transmission rate can be faster and can be slower.
In my line of work, many of my users have the same kind of misconception as Alfred. They think that when I say the throughput to the company’s server is 70Mbps, they are able to get 70Mbps all the time. This is a concept that I have to explain to my users again and again.
I would also like to point out that Alfred’s suggestion of listing the throughput to the top 10 sites as a gauge for real world experience is misleading. These websites are served by CDNs like Akamai and are often cached by proxy servers. Throughput to these sites are naturally high for all ISPs and will not help us differentiate between the ISPs.
The bottom line is consumers need to realize that ISPs are really trying to make a profit to pay their employees and stock holders. There’s always some “marketing” involved to make consumers feel that they are getting their money’s worth (you’re never getting your money’s worth in consumerism).
Consumer’s 100Mbps is not real dedicated speed. They are *shared*. That’s how money is made, just like the insurance business. “Up to 100Mbps” sounds much better than than “not 100Mbps most of the time”.
I believe that forcing ISPs to disclose these “truths” is just a futile attempt at disrupting their business models. Unhappiness stem from the fact that there are some users (the demanding 20/80 highlighted by Alfred) who wants/needs/desire absolute 100Mbps.
ISPs and IDA could learn from the Singapore public transport model. If you’re poor or just want spend your spare change, be happy with taking the high latency and crowded bus or train. If want your own fast dedicated point to point transport, pay more for a cab. For the impatient I-want-it-right-now guy, feel free to call a cab or even ask for a Mercedes.
Thus for the 20% demanding users and pirates with need for speed, ISPs could just simply create a premium tier for them. Instead of “up to 100Mbps”, just pay 5 times more for a “dedicated 100Mbps” line.
Money talks, so if you want more, pay more. IDA should just force the real premium price plans out of the ISPs instead of forcing fluffy average speed out of the low end plans.
If you’re late for work, pay to take the cab. If you need to watch latest pirated movie immediately, pay 100Mbps premium.
Sorry, noted that you revised to 16mbit US average later on, rather than 25 as I wrote above. 🙂
For clarity’s sake, I do indeed subscribe (unhappily, overall) to StarHub’s 100mbit plan, and I can assure you that it does actually reach peak speed (I can get 7-8MB/sec on encrypted/non-shaped traffic from well internetworked sites). However, you’ve probably right in your revised figures about average speed being 25mbit regardless. Having said this, it certainly does not follow that because their average speed doesn’t go higher than 25 that it’s not “worth it” to get a 100mbit connection. As an individual I may value getting that occasional 100mbit burst, if it happens at the right time of day for my usage profile. That’s an individual decision.
Having said that, I’d happily trade my 100mbit down to even 8mbit for better latency and consistency. For this reason I will take the anecdotal evidence I hear about Singtel’s superiority in this regard and probably tell StarHub to shove it when my contractual period is up.
I can also assure you that Starhub, over the last 9 months or so, has started capping speeds at 10mbit after an unspecified (and varying) data cap is hit. I don’t know if this applies to all plans, or just the 100mbit one. While I don’t philosophically disagree with data caps as a way to manage behaviour that might be anti-social to the network, I DO disagree with un-advertised, un-communicated ones that completely lack any transparency and fail to set any expectations of customers – that helpdesk staff have no idea about. How do I know what level of data is expected of “reasonable use”? I also disagree with the crude throughput limit they seem to have implemented when one hits these caps. They seem to be doing it in such a way that causes hell with TCP connections and basically makes the connection unusable when anything is downloading while capped.
I’m not sure if anyone reading has heard more than I about this rate limiting policy, but I haven’t been able to find any communication from StarHub on their website at all. Seems amiss to me….
Hey Chad,
Glad you pointed out this throttling/ rate limiting/ traffic shaping policy that telcos apply. It’s something that doesn’t seem too transparent at the moment, which is rightly making many users unhappy. Both sides to the arguments apply, I feel.
As profit-making companies, telcos do have a right to design their networks to be optimised in a way that both makes money for them and satisfies most customers – often the saying here is 80/20, 80 per cent use of folks use only 20 per cent of the bandwidth, so overall, telcos do make money while keeping most users happy.
However, when they throttle traffic (likely Bit-torrent traffic, for example) as aggressively they are known to do, I agree with you – IDA may want to get them to be open about it. If BT is supposed to be something they think should legitimately throttle, then they should be open about it. Reason why they haven’t done so could be because they still want to sell the high-end plans to heavy downloaders (who do get the max speeds through aggregation, ie concurrent downloads from different sites).
Another poster AY suggested telcos should also reveal their over-subscription rates (ie how many people are made to share X amount of bandwidth), so that they will not cramp so many users together to share a pipe which results in the slow speeds during peak hours around midnight, when everyone goes online at the same time.
Unfortunately, that’s a lot of info which consumers may or may not understand, plus it may even make it impossible for companies to compete if they reveal their over-subscription plans (in essence their business plan). So, I feel the IDA has to balance the need for transparency to consumers and the ability of telcos to compete.
No easy answers here, certainly not with a simple “average score”, as you rightly said earlier. My personal suggestion would be to go with something more detailed – say, a real-world snapshot of the performance to the most popular online destinations here, during peak and non-peak hours – and for IDA to mandate that such figures are published. Not sure if that will be perfect, but at least it seems like a “real world” speed that many users can relate to.
Wow… Epic thread. Took me a while to read through everything.
Actually, I like the examples CK provided from IDA. I didn’t go and double check his figures, but looks correct. Quite shocking that Starhub is shortchanging its subscribers. I wonder what will happen if Starhub is forced to published those figures.
As for CK being insulting to everyone who disagree with him, LOL, sounds to me he is only insulting Alfred.
Anyway, peace to all. Happy New Year.
Great article Alfred. And great discussion. Sorry CK, I choose to agree with Alfred more.
I think another factor that will affect your broadband speed is the number of people using the same ISP in your area. The average speed of someone in AMK will be different from someone living in Pasir Ris. And the time you access internet will also make a difference. I got great internet speed between 4am to 6am. So how can we make ISP release a meaningful “average speed”? What exactly is average speed? How do we measure average?
I think the current model of advertising the top speed is working out fine. IDA should look into quality assurance of the broadband speed.
For readers who are following this thread:
After reading through this today again, I feel it is important that we be clear why one of CK’s comments was removed, by showing it in full.
Since he has apologised, I shall leave my comments aside and let readers see for themselves how this discussion has unfortunately descended into condescending remarks made by CK.
His charge is Techgoondu has “censored” his post and thus made this an unfair discussion, so we have decided to show how he has taken this discussion out of context and repeatedly insulted anyone who disagreed with him that they do not understand the simple math of averages.
He has been warned in the previous posts but unfortunately, has repeated this again in the deleted post.
He subsequently edited this post (please see above), which we did not hesitate to post in full. Below, however, is the original post, which repeatedly insinuated that anyone who disagrees with him is unable to understand the basics of averages.
As is clear here, this is not a question of Techgoondu censoring remarks on its site, which we have never done prior to this, nor is it because we do not accept criticism as we level them ourselves at times.
Rather, this is a question of keeping our discussions clean from personal insults, and making sure they are useful for our readers. It should be clear from the following:
You said:
“If you print an average speed 25Mbps on your brochure, does that mean I will get 25Mbps to the sites I go to? If the answer is no, then how is that a real gauge of what I will get?”
The answer is NO! You really DON’T understand what is average speed do you? If I can get 25 Mbps to [every] site, it is the MINIMUM GUARANTEE SPEED. NOT AVERAGE SPEED.
If you have an average speed of 25Mbps, you could have real world constant 25Mbps. But it also means that you can have some very fast speed exceeding 25Mbps, and some slow speed well below 25Mbps. But on the AVERAGE, it is 25Mbps.
If I tell you the average speed on the CTE now if 60km/hr, do you expect all the cars to travel from the start of the CTE to its end at 60km/hr?
I would think not. Cars in the fast lanes will go faster, cars in the slower lanes will be slower. Some more congested part of the CTE will be slower than the more empty parts.
But is average speed misleading? I don’t think so. I would use it as a gauge to whether to use the CTE now or not. If the average speed now is 90km/hr it means CTE is a good path home. If the average speed for CTE is now 40km/hr, it is not a good path.
Now.. I don’t think other people don’t understand average. I think YOU don’t understand average.
Okay, let’s leave it at that. This is not a wasted discussion. It just needs to be focused and not descend into personal insults. Apology accepted and let’s keep the thread open for other readers.
“If you print an average speed 25Mbps on your brochure, does that mean I will get 25Mbps to the sites I go to? If the answer is no, then how is that a real gauge of what I will get?”
The answer to your first question is NO. If you want to get 25Mbps to every site, you are asking for for “minimum guaranteed speed” not average speed.
An average speed of 25 Mbps can be constant 25 Mbps to all sites or it can be much faster speed for some sites, with slower speed for others, but on the AVERAGE, it is 25Mbps. This I feel is the source of our disagreement.
I want to try to repost the above paragraph because I think it is important to the discussion. This is the final posting I will make on this site. In a situation where you can keep accusing me that I did not get my facts right and I do not understand the topic I am saying and yet when I point out your mistakes, you choose to use your powers to delete my comments, it is no longer a debate on equal grounds.
I have strive to use the facts you have provided and try to answer the questions you posed to me diligently, while you have not even attempt to answer the questions I have posed to you repeatedly and refuse to provide your interpretation of the scenarios that I have constructed.
This has been a wasted discussion. I should not have started it in the first place. Finally, you keep saying I have insulted you, while I fail to see how, let me offer my apologies to you. Please do not take it personally.
CK, I will approve of comments that are beneficial to readers, rather than ones which trade in insults.
I have given you the same right of reply as all our newsmakers and have only removed your comment that was particularly insulting. I leave the rest of the comments here to let the reader decide what is going on during this thread.
You were given the last say on the issue. You chose to repeat an insult you had posted earlier in this thread, which I’m sorry, does not benefit our readers who are following this.
The thread is still open. People can still comment. But just like on any thread, any offensive comments or personal attacks are removed.
I am disappointed that you choose to censor my comments.
I am merely pointing out what you have misinterpreted and I think it is important for the readers to understand the basic flaws in your argument. If you truly want to let the user decide, you should not stop me from pointing out the problems in your arguments.
IF you choose instead to prevent me from making my case, it would seems that I have wasted my time here today as you are not courageous enough to admit you are wrong and choose to use your power over the commenting system to make it look like I have insulted you.
You said “CK, I leave it to the reader to decide if your figures are anywhere close to what they get on a daily basis. I take that to be your last word on things.”
Yes indeed. I think we should leave it to the readers to interpret for themselves. Now I see it clear and wholeheartedly agree with what you said: “It’s also about educating users what they can, and cannot expect.”
You can make all the data available to the people, but if they cannot understand and make use of it, it is ultimately useless.
CK, I think you’ve made your point over 34 comments and there is no need to trade insults in a public discussion. That is the reason why your last comment has been removed.
CK, I leave it to the reader to decide if your figures are anywhere close to what they get on a daily basis. I take that to be your last word on things.
Meanwhile, I’m leaving this thread open for other readers to comment on the issue.
Okay… I have made a mistake. I took the wrong values. I have mistaken the local throughput for overseas throughput.
Here are the correct values for overseas to US throughput:
SingTel 10Mbps plan: overseas average speed for the Sep-Nov: 9.477 Mbps
Starhub 30Mbps plan: overseas average speed for the Sep-Nov: 16.126 Mbps
M1 15Mbps plan: overseas average speed for the Sep-Nov: 9.2047 Mbps
This is even better for our analysis. Firstly, Singtel’s 10Mbps is closest to the advertised top speed. Secondly, you are like to get only slightly more than 1/2 the Starhub advertised top speed. If you combine this results with the local throughput results, Starhub cannot even meet its local speed commitment! M1’s average throughput is only about 9Mbps.
So my conclusion is that Singtel’s 10Mbps plan is closest to their advertised top speed. For M1, it is 10Mbps.
If you work out the average speed for all the plans for all the ISP, you will realize that Starhub is especially bad:
Its 30Mbps only really provide 16Mbps on the average. What is worse is their 16Mbps provides only 10Mbps on the average, although they are fully capable of 16Mbps as evident in their higher level plans. They are clearly apply unfair network management practices to their customers.
Anyway, thank you pointing out the information on IDA’s website. I think it will be good if these information is made available to as many people as possible.
Same question CK.
If you print an average speed 25Mbps on your brochure, does that mean I will get 25Mbps to the sites I go to? If the answer is no, then how is that a real gauge of what I will get?
Think about that and answer the questions truthfully. I will take your last word in this debate, because I think your points are woefully short and not benefiting readers anymore.
Let me sum your points up:
1. You say average speed is good but yet you have to ask your friends what is really good.
2. You say people who disagree with you do not understand what an average is, yet you do not answer if the average score is something you can get in reality.
3. You bring in Net neutrality, yet do not understand that the debate is not about having the same bandwidth to every single site but about people against ISPs charging more for content.
I think first you may have to convince yourself of your logic in these arguments first before attempting to persuade others.
Oops wrong symbol:
Top Speed > Average Speed = NO GOOD.
LOL. If you still don’t get it, then I have nothing else to offer.
In the example I gave, Starhub’s 30Mbps plan gives average speed of only 25Mbps over the months of Sept to Nov. This simply means that if you are a user of Starhub’s 30Mbps plan for Sept to Nov, you are getting likely you get less than what you have paid for.
To put in simpler terms:
Top Speed = 30Mbps
Average Speed = 25 Mbps
Top Speed < Average Speed = NO GOOD.
This is the best example why average speed is a much better value than top speed. But yes! Let the other users judge for themselves.
As a footnote:
You might or not realize it. The term throughput itself is an average. It is a average of the amount of data moving between two points over time. Which is why the unit is Megabit PER second. Thus, like all averages, it is dependent on the sample population, in this case, the time period over which it is calculated.
Same question CK.
How does an average speed inform people better if it is not what you get in reality? If you apply the logic to unreal top speeds, try applying it here.
If this is just a gauge as you say, it is no more a gauge than top speeds. At least you do get that.
CK,
Which fact of life would that be? I’ll let readers look at your posts in full and decide for themselves.
And these numbers you have derived from the IDA site, I wonder what they inform of the services. I won’t guess at the intention. Again, I’ll let readers following this debate decide for themselves if they understand what you are arguing about.
My answer has always been the same: You use it as a gauge, an indicative value.
In the example that I have provided, I would not subscribe to Starhub’s 30Mbps or any of their higher plans because it is clear that Starhub’s bandwidth is limited to 25Mbps on the average. Of course this will change over time, but at this very moment, Starhub’s 30Mbps plan is not a good plan to go for.
Would you interpret it differently?
Finally, I am not trying to insult anyone. I am merely pointing out a fact of life. While most people can perform averaging, not everyone can interpret the values of averages.
Take for example people who buy financial products based on its average profit for the past 3 years. Or buying stocks based on how the STI is performing.
Statistics is a powerful tool, but only if you can read past the values and interpret the results correctly.
Let me help you along by using the values from IDA website for each of their highest plans:
SingTel 10Mbps plan: overseas average speed for the Sep-Nov: 10.644 Mbps
Starhub 30Mbps plan: overseas average speed for the Sep-Nov: 25.180 Mbps
M1 15Mbps plan: overseas average speed for the Sep-Nov: 15.479 Mbps
I have taken these values from IDA website and average them out. I would like to hear your interpretation of these values if they are mandatory figures in the sales brochures.
Firstly, I would like you to address the examples that I have listed and tell me why average speed is not helpful or misleading in those context that I have provided. Am I understanding it the wrong way?
Secondly, stop taking my comments out of context. I said:
“Conversely, if there is a particular unpopular website that I need to visit regularly, it makes logical sense that I test it with friend’s connection first, because it is not indicative by averages.”
I am merely trying to illustrate how average would not help at all in certain situation. Why may I ask is that a bad thing? So far, I have seen you keep saying average speed will not be a good thing. So is it a bad thing in every situations?
AL said:
“1. Average upstream International bandwidth allocated for consumers and business customers
2. Average latencies provided from these upstream ISPs
3. Over-subscription ratio of bandwidth to consumer and business customers”
and I said:
“Once you are on plan that is higher than the average speed of the external links, the top speed no longer matters, because you are limited by these external links. So in a way, average speed is a representation of these external links once you are limited.”
I am agreeing with AL that once your local bandwidth is above those of the average upstream International bandwidth, it is the latter that really matters.
Finally I said:
“If youtube and my web server are both about to provide 10Mbps to every user, that is what the user should get and not getting better youtube connection because my ISP paid for better connection to youtube because it is more advantageous for its marketing.”
This is to say that if any two service on the internet is able to provide the same bandwidth for their service to users in Singapore, they should be allow to do it, and not limited by the deals that are cut by our ISPs because it is advantageous for their marketing. I do not know how you interpret that as that I am “asking: 10Mbps to every site”
Lastly, can you provide an example of how average speed is misleading to you if this information is presented to you in a brochure from the all three ISPs? I have presented examples how I would use the information. I would like to see how you will use it.
I think first of all, CK, look at the contradictory statements you have made and try convincing yourself first.
My question to you remains: how does an average number inform users better if users cannot get that speed?
Answer that and you will see my point.
Finally, I think you owe me and the other folks an apology for insinuating we do not know simple math. We do. Just that we disagree with your flawed logic, which is obvious to all here.
CK,
I believe you have to review some of the things you have said on this thread and see if you are convinced of them yourself.
You said: “It seems to be you belong to the category of people who do not know how to use it properly. Apparently, there will always be people like you who will keep complaining, so there is no stopping it.”
You also said: “While it will not help people who cannot understand what is average and how to interpret it correctly, I for one, welcome the additional information.”
I think there is no need to insult others by implying they do not know the simple math of averages. People here (including the other two folks posting here) are questioning the practicality of the scheme you are advocating – they are not unable to understand your idea, nor are they so unintelligent to not know maths, as you suggest.
Don’t forget what you said yourself: you have to actually ask a friend for his experience even with this score you are proposing. How different is that from what we have now, how much more “real” is this figure and how does it benefit the consumer?
If you can answer those questions convincingly while being honest with yourself, good for you.
As for Net neutrality, I’m not sure where you got the idea that this was an “overseas issue”. I was pointing out the background to the debate which has emerged overseas, which was about telcos trying to charge more for certain services. People are against that, but they are not asking for what you are asking: 10Mbps to every site. It’s a simple difference you can surely see.
Like you, all consumers want more competition for our dollar. We seek transparency and details. But we seek the right details that give us a better buying decision, not another guess-timate. And that’s the crux of the issue here.
Well, thank you for taking your time to “correct” me.
I have listed out ways how to use the average figure and how not to use the average figures. It seems to be you belong to the category of people who do not know how to use it properly. Apparently, there will always be people like you who will keep complaining, so there is no stopping it.
The one thing I can agree with you is that IDA should implement this scheme. While it will not help people who cannot understand what is average and how to interpret it correctly, I for one, welcome the additional information.
The crux of the whole exercise is that IDA wants the ISPs to indicate it in their marketing, so that the consumers can have access to it without doing research. This will hopefully spur competition which is the important thing here, not to shut up whiners.
As for the network neutrality issue, it is typical of Singaporeans to think it is an “overseas issue”. If you understand how the inter-connected networks work, you will know that it will affect us and our businesses in more ways then just an “overseas issue”. But since you already come to a conclusion that I do not understand the concepts of how the content delivery systems, peering arrangements, transit policies etc work, I will leave it at that.
Happy New Year! Thanks for your time.
CK, thanks for the debate. I hope readers here will get something out our different views. But please allow me to correct some of your points.
The whole exercise of putting an average speed is to make sure that it is a “real world” speed, but obviously, it is nowhere more indicative than what the top speed offers. You admit that as much by saying you gotta ask your friends what their experience is. What does this bring us back to? Another number that is not indicative?
This is not about me defining what average means. It’s about how it benefits consumers. Look at the post above yours (Chad). There are many issues that affect users here, and simply publishing an average speed number not only doesn’t solve the problem. It confuses.
I think you also misunderstood my point about asking for more details. Consumers deserve more details – and there are already a lot on the IDA site, trust me, take a look there – but these have to be the right details, not another fuzzy figure that doesn’t deliver what’s on paper.
You have also not correctly addressed a lot of concepts about how the Net works, like Net neutrality and the idea of inter-connected networks.
The whole Net neutrality debate was sparked because ISPs overseas were threatening to charge more for some websites that are popular, which people were up in arms against. People are NOT asking that ISPs give 10Mbps for every site. You have to get that right first.
You’ll understand that if you also understand how the Internet works. It’s not a straight road that one ISP or one content provider offers. It’s a lot of inter-connected networks that not a single ISP controls. As soon as you understand that, you’ll realise there are many more issues out there than simply having an average speed figure published.
The best test, I guess, is to have IDA implement this scheme. Let’s come back after that and see if people have stopped complaining that their ISPs don’t give them the speeds they paid for. 🙂
My friend just point this out to me:
While you are talking about average speed, in your mind, you are really thinking about guaranteed minimum speed. That is why you are always talking actual speed, how it is impossible to ensure high speed to all websites and how consumer will be confused and disappointed by “average speed”.
Okay. I think we can stop now. I think my deliberate example has served its purpose to bring out the differences in our thinking.
Clearly your understanding of averages differ from my mine. For me, averages works because you are talking about the popular stuff. If a thing is popular, it affects the average. If it is not popular, it will have little effect on the average. As such, it is indicative.
For you, as long as one of them is not the norm, average speed has failed to serve its purpose. For you, average speed doesn’t work because if you surf to a website that is far below the average speed, it is the ISP’s fault, without understanding what the word average actually means. This is the kind of thinking that requires education as analysis of the additional data provided has clearly failed and you are drawing the wrong conclusions.
Likewise, your idea of network neutrality is all about money. Yes money is part of network neutrality. But more importantly for me, network neutrality is about equal opportunities that the Internet has come to stand for. If youtube and my web server are both about to provide 10Mbps to every user, that is what the user should get and not getting better youtube connection because my ISP paid for better connection to youtube because it is more advantageous for its marketing.
Let’s try another situation.
For example, if a new service springs up in Vietnam call metube.com which is gaining popularity in Singapore. It is popular, but it does not make the top 10 popular site. Let’s say Starhub notice the trend and arrange to get better connection to Vietnam. Its average speed will increase as more and more people surf to metube.com while the other ISP average speed begin to drop as metube.com gain popularity.
Of course this is a gross simplification of the actual deals that the ISP actually does behind the scene, but at least the ISP is now motivated to improve their links rather than not having to do anything at all.
Here is another one:
If Starhub keep signing up new customers without increasing their overseas bandwidth, their average speed will start to drop over time. Now, imagine that your contact with Starhub just ended. Do you really want to sign up with Starhub again, or move to another ISP with better average speed?
It might not be the perfect indicative metric, but at least it is something that will promote competition. And of course the ISP will not want to have something like that.
Let’s see now:
If Singtel provides an average speed of 10Mbps, Starhub provides average speed of 8Mbps and M1 provides an average speed of 9Mbps,
Which one would you prefer to sign up?
Now given that Singtel provides 8Mbps, 16Mbps and 100Mbps plans and you only surf mainly popular overseas websites, and want the fastest speed without paying too much for it, which on makes the most sense to go for?
I would chose Singtel 16Mbps logically. The additional information is useful isn’t it?
Conversely, if there is a particular unpopular website that I need to visit regularly, it makes logical sense that I test it with friend’s connection first, because it is not indicative by averages.
Now if you look at the current system of top speed. How do you draw any useful conclusion from it?
I really don’t get why you keep harping on how the consumer will feel getting cheated when they surf to a slow site. If the site is slow, EVERYONE feels it is slow.
Bottom line is this. While I don’t think average speed is great, I think it is much better than the current top speed centric marketing and I for one, will welcome the change.
And yes, ISP do prioritize traffic now. But your idea of a published list of websites disadvantage them in bargin power. And it doesn’t reflect underlining changes to internet changes. In the end, it is far more misleading than average speed.
CK,
I think you answered your own question there. If average speeds give me a real reflection of things, why do I still need to check with my friends?
As for Net neutrality, I will leave it to you to reflect on my points. When people argue against Net neutrality, it’s not that all content get the same priority. It’s arguing against consumers having to pay extra. You have to get that right. Prioritising traffic is a different issue.
I think AY got it correct.
It is the external links that really matters. Not the published local speed. While top speed matters (as it has to be sufficiently high), it should no longer be the thing that the local ISP be competing on.
Once you are on plan that is higher than the average speed of the external links, the top speed no longer matters, because you are limited by these external links. So in a way, average speed is a representation of these external links once you are limited.
And precisely because consumer plans are “spare change” to ISP, the only way we can get better service is if we are empowered with information that will enable us to make decisions to migrate from a lousier ISP to another en mass. Only then are ISP forced to improve their service.
Btw, also thanks for reading our other articles!
The point about Net neutrality. I think you’ve made a wrong comparison with pay-TV content. Pay-TV is expensive because they are exclusive. There is no exclusivity on the Net, even if ISPs decide to prioritse traffic to certain sites.
And let’s admit it. ISPs DO prioritise traffic to certain sites all the time. When it comes to the Net neutrality debate, it’s not that we want them not to prioritise but that we do not want to pay extra for such content.
Asking ISPs to increase bandwidth to the most popular sites is far from asking them to go into exclusive pay-TV deals.
@CK: Let’s agree to disagree.
To be honest, you haven’t answered the questions. If this average speed is supposed to be a better gauge of reality, then what will the consumer think if he STILL cannot get that average speed when he, oh, surfs to a slow site? How does that illuminate his buying decision better than a top-speed rating?
If we want people to get what they see on paper, then I don’t see how an average score does any better? In fact, it confuses.
Secondly, thanks for agreeing that average speeds are very far from the perfect gauge. I think that was my point. You say top speeds have “ZERO” bearing on reality. Try downloading from one of the sites I mentioned and you’ll be surprised you can often get pretty close to what is on offer, because these sites do not throttle and they are local.
I think we both agree that things have to be real, not theoretical. I just do not see the point in pasting an average score which does not make any more sense than a “top speed” rating now.
I am not asking fewer details – I’m asking for the right details.
The poster above you posed a few interesting details that IDA can get ISPs to disclose. That is what people should look for, not an average score that doesn’t solve any of the problems plaguing users here for years.
1) It benefits the consumer in two ways. Firstly, it is a changing value based on actual usage rather than a theoretical published value that is never achieved in reality. That alone makes it much better than top speed. Secondly, the ISP will actually need to work hard to keep their values up. The situation now is that the ISP just have to deploy some hardware on their network and then they can claim that they provide UP TO X Mbps of bandwidth without improving the actual outside links.
2) They can’t. Which is why average is being proposed. I think you still haven’t actually sit down to analyze how average works. Average take into consideration both top speed and the popularity of the sites. It is very far from being a perfect gauge. But it is MUCH better than the current top speed advertising, which has ZERO relation to reality. At the very least, please take some time to think this through.
As to why shouldn’t ISPs be put into a position that they have to pay more for certain network links, you just have to read your own article about the TV telecast of soccer matches in Singapore. It is precisely that Starhub and Singtel are in a extremely bad position to bargain, that we consumer end up paying much more.
The POINT of IDA’s proposal is to give consumer a better metric based on actual usage rather than theoretical top speed which is TOTALLY useless. It is NOT a representation of the actual experience, BUT it is much better than the current one as it discourage the ISP to simply improve on the local network speed (and thus using it for advertising) without improving the more important overseas links.
It is also a metric that the ISPs need to WORK in order to improve on, not something that they can just claim to have simply by changing equipment. This is what is good for competition.
Lastly, any priority of any network destinations over others is bad for net neutrality. By having IDA endorsed certain list of sites, we as consumers and businesses stand to lose as we cannot compete for traffic with well established destinations.
I believe all these debates are useless you’ve worked in an ISP and understand the perspective and business plans on how to run an ISP. Don’t ever forget these facts:
1. ISP is a FOR-PROFIT business
2. Your $80/month CONSUMER plan is *spare change*
3. Internet = Inter-connected networks of other FOR PROFIT ISPs
The combination of inter-connected networks of different entities with different business priorities, plus bandwidth and latency limitation makes it hard to commit to any fluid average numbers.
I agree as a consumer, I would much prefer my 16Mbps plan to be as true as possible to every single service that is important to me and me alone. However I would make a very narrow and uninformed argument to shout about my narrow minded desires without looking at the big picture.
You can only make a valid argument about “speed” if you have an ISP’s business plan on your hand and if you’ve seen upstream provider contract.
I agree that IDA is barking up the wrong tree. Either they don’t understand how ISP business work or they are just caving in to petty and narrow minded consumer complains. Instead they should push for the following transparency from individual ISPs:
1. Average upstream International bandwidth allocated for consumers and business customers
2. Average latencies provided from these upstream ISPs
3. Over-subscription ratio of bandwidth to consumer and business customers
These facts above are what ISPs can control and they will directly impact the final average performance regardless if you’re just a web browser or P2P pirate.
@Chad, yes, agreed there. The lack of consistency is a bigger issue that “average speeds” won’t solve.
Singapore users surf to tens of thousands of sites – this average speed is simply an average of too many disparate sites and don’t give a true reflection of what is real usage, just like top speeds don’t.
The difference is at least for top speed, you can attain it if you go to a local site. Do you actually get an average speed when you surf?
As for consistency, I think that’s where the IDA can be more stringent. Make telcos report what their latency and throughput are on peak and non-peak hours (please see the IDA site where they report such figures), so that people can see how consistent a telco is.
I’m afraid there might not be space for all these charts in telco advertisements. But what IDA can do is to mandate that ISPs provide information on how to access these reports on its site.
Also, IDA may need to make the information more easily understandable than it is now, if it wants to educate the masses because Throughput and latency don’t mean anything to many broadband users.
@CK, I’m afraid you’re missing the points mentioned in the article. We could move the debate forward if you could answer these practical questions:
1) How can average speeds benefit users by giving them a better clue about what they will get in real usage? (They cannot because everyone surfs to different sites, and I WILL NOT get the same “average speed” on paper, just like I won’t get the the same top speed on paper)
2) How can a telco give you good speeds to all sites (they cannot, because there are tens of thousands of sites and many of them throttle, shape traffic as and when they want)?
If we cannot come up with any good answers to these questions, then I’m afraid everything else is academic.
As for the top 10 sites, I take your point. They need not be “sites” per se, but online destinations – that should include your non-HTTP traffic. Add a WoW server, if it’s among the top 10 destinations.
Would this make content providers force ISPs to pay more? Perhaps. But why not, if these are the sites that Singapore users go to? Conversely, these content folks are bringing their content worldwide, in a way of reaching their audiences via distributors like Akamai, so, to simply say that ISPs will end up paying more is just looking at things one-sided.
As to your highway analogy, I’ll ask you this: has a Singapore ISP or any ISP, for that matter, ever promised a 100-lane highway to everywhere on the Net? Does any ISP in the world do that? I’m afraid not.
Even if you sign up for a 10Mbps leased line with an SLA, you know you get router-to-router speeds, not 10Mbps to everywhere.
So the point of this debate is not just telling telcos to be transparent. It’s also about educating users what they can, and cannot expect. That confusion has been around ever since the first time I hooked up a trial cable modem in 1998, and it is still around today, because people somehow think that the Internet is as simple as the telco upping speeds to increase bandwidth.
I’m for transparency and I’m for more competition. But first, people need to stop barking up the wrong tree and blaming telcos for everything. Have you thought of why Tudou.com or some overseas sites are slow, by the way? It’s also because they limit how much we can download, depending on their server loads. That’s the question that people don’t answer, don’t want to answer.
If anything, look at the throughput speeds at the IDA website. There’s a ton of info there – not perfect, no – but folks who want to complain about a lack of info should go there and have a look first.
I think it is also important to note that HTTP traffic just a part of the overall network traffic. FTP, VoIP, IPTV, gaming, video and music streaming are just some examples of traffics that are NOT HTTP based. Having top 10 sites also prioritize HTTP over all other forms of internet traffic.
I also think it is important to examine where IDA is coming from. Take Starhub as an example. Year after year, the advertised speed of its MaxOnline has been ever increasing. But if you can ask any MaxOnline user, has the overall experience been increasing as advertised.
While tech savvy people like us are in a position to judge for ourselves whether we will pay the extra for the incremental advertised speed of the higher tier package, most consumers cannot understand the differences between top speed and actual speed. As such many end up paying for a more expensive package while getting an experience much lower than expected.
To use my analogy of the highway, it is like the ISP telling the users what they have 100 lanes highway, but neglecting to tell them that there are 1000 vehicles on the road with you, and the roads to the destinations you want to go will only have 5 lanes because the ISP did not pay for better connecting roads with 30 lanes.
Please don’t jump all over the place. IDA is not asking ISP to guarantee minimum or average speed. This is not what IDA ask for, neither is it what I am talking about. The idea is to force the ISP to publish the values so as to spur competition by giving consumer additional information to select their ISP. And there is no guarantee in any of the consumer contracts, so how is this ever a problem?
Precisely because it is impossible to offer perfect connections to all website, _average_ speed is a much better measurement. If you understand how average works, if there are a lot of people surfing to a set of websites with few people surfing to less popular sites, the average speed will be high if the ISP’s connection to the popular websites is good. But once more and more people surf to sites that are not connected well by the ISP, the average speeds drops. It is then a good indication that the ISP has not been improving their connections sufficiently.
If you have been following the net neutrality debate at all, you will know that having top 10 sites is a bad idea because it prioritize traffic of popular sites over non-popular sites. ISP is encouraged to do their very best to keep the connections to these popular sites as high as possible while they can neglect the less popular sites never going to ever make top 10. In this way, if I surf to less popular sites regularly, I will never get good access ever, because I don’t have the same surfing habits of the masses. This is also bad for small businesses or startup, because the ISP is not motivated to provide good connections to their web servers.
Besides, if there is a list of pre-determined sites, CDN providers and net traffic providers will be in a much better position to negotiate with the ISP because they know that the ISP will need to have good connection these sites for this month. This will inevitably drive up the prices the ISP will have to pay.
I think your commentary would be more complete if you also addressed the lack of discussion about overall “quality” of connectivity including latency, DNS reliability. While speed is important for media-heavy sites, latency (and consistency of latency) is a major factor in making browsing “feel” fast to the end user.
It’s also part of the debate completely missing from the mainstream media (unsurprisingly).
The /quality/ of the internet experience here is what is disturbing to me. While the headline, and actual, broadband speeds I get here are amazing at the right time of day, and right night – I find it incredibly unreliable compared to my experiences in other countries (New Zealand & Australia mainly). Sites randomly time out, DNS is slow, clicks often feel like they take eons. Low res youtube videos stop working halfway through. Online gaming can be a crap shoot as a result.
I don’t know what causes it – traffic shaping and packet inspection? great firewalls? crappy proxies? poor links or congestion in HK or JP? But it certainly merits discussion to me.
I certainly agree with you though that regulation of advertised speeds will do more to confuse the issue than to improve the average customer experience.
CK, as a consumer, I agree with some of your ideas in theory, but in practice, I gotta be realistic and say some of them don’t work.
First, how useful is an “average” speed to users when we all surf to different sites, i.e. if Telco A says it offers 1Mbps while Telco B offers 1.4Mbps, does that mean I can really get faster speeds with Telco B? If it doesn’t, then this brings us back to the original situation – I don’t really get what I see on paper.
Second point: how can telcos guarantee minimum or average speeds to all sites? I think that’s the question that no one wants to answer, because deep inside, even though we want that guarantee, it’s not possible to have for the $40 – $100 we pay for consumer broadband.
There are service level agreements and bandwidth guarantees in the corporate world, where telcos have to pay out damages if they do not keep to bandwidth requirements of their clients. But those costs tens of thousands of dollars a month and even then, these are often guarantees of how big a pipe is offered. No telco will be able to guarantee a speed to every site you surf to.
That’s reality, unfortunately. And the reason why I suggested 10 top sites is because it gives a snapshot of what the telco provides to MOST users. It’s important that the information gives some idea of what users can expect, but it is just as important to inform users what they cannot expect – that is speed guarantees to every site.
This top 10 list can change from month to month, so telcos can “game” all they want, but at least they would be providing the bandwidth that most users want.
Compare this to an average bandwidth “score” which includes sites where few of us visit, or where the site owners do the throttling of downloads. Is this fair to that telco which suffers a slower score as a result? More importantly, is it instructive to most of us who don’t visit such sites?
I think your analogy is totally wrong. Cars are more like the routers and modems that user uses. ISPs are not car manufacturers. They are road builders who build roads and collect toll fee from road users. After some times they stop expanding and maintaining their roads, and yet allow more and more users onto the same old road, you are just going to get traffic congestion. Think more of CTE and the ERP as analogies.
“Should telcos advertise an “average” speed based on where its users surf to each month, considering that this changes because users surf to different places offering varying download speeds each month?”
Yes! If most users go to certain websites, the average speed will be high. However, if more users start to surf to different locations, thus dragging the average speed down, then it is a good indication to the user that the ISP does not provide adequate links to alternative sites that are rising in popularity. And it also serve as a warning to the ISP that they should improve the connections to these sites in order to raise their average speed.
“Instead of a meaningless figure, the IDA should get telcos to show how fast they link up to, say, the top 10 sites that users here go to. If these throughput tests are run by IDA, they should give a transparent, unbiased view of which telco is giving users the best bang for buck.”
If however, the speed published is restricted to only a pre-determined list of top 10 sites, then it is easy for the ISP to negotiate priority access to these sites while neglecting the connections to other destinations. This is detrimental to the users as internet tends changes much faster than IDA can update its lists and will end up misleading the users with outdated data. The important thing is to keep the ISPs on their toes and not be complacent.
However, all these information is useless unless the consumer can act upon it. LTA should make it easy for the consumer to break the contact they sign with ISPs if the ISPs are unable to keep up their promised standard of service.
It is also important for ISP to be transparent about the kind of traffic management that they impose on their users, so that they can make good and informed decisions.