Not so fast.

All batsmen have struggled to get a start at some time
Every batsman is vulnerable when they first go in, so all batsmen have occasions where they get out early in their innings. These poor starts will sometimes group together because that's the nature of random events.

A good batsman will make the best of his starts
The batsman you don't want is the one who gets out after he's got himself in. Let's look at what Lyth has done when he's faced more than 3 overs:

• New Zealand, 2nd test (Headingley): 107 runs before being cruelly run out by Gary Ballance
• Australia, 1st test (Cardiff): 37 runs, miraculous catch by Clarke at slip
• Australia, 3rd test (Edgbaston): 1st inns, 10 runs, standard Lyth dismissal in the slips
• Australia, 3rd test (Edgbaston): 2nd inns, 12 runs, LBW Hazlewood
• Australia, 4th test (Trent Bridge): 14 runs, standard Lyth dismissal
• Australia, 5th test (Oval): 1st inns, 19 runs, terrible pull shot off Siddle
• Australia, 5th test (Oval): 2nd inns, 10 runs, standard Lyth dismissal

So as the pressure mounted after his early bad luck, Lyth tended to get out to his usual faults outside the off stump even when he'd got a start. The problem now is how to manage that pressure as well as correct the technical fault that leads him to play at balls he could leave.

Lyth's strike rate declined as the summer progressed. This mirrors Michael Carberry's experience in Australia where he arrived in an aggressive mood and became effectively shotless as the tests went on. It's a response to pressure: not getting out becomes the driving force rather than scoring when the opportunity presents itself.

So who else can partner Cook?
Ah.

This is Lyth's biggest asset at the moment: the lack of alternatives. None of Cook's previous partners have scored heavily since being dumped. Carberry has come back into some form recently, Compton is showing signs of resurrection, but neither is demanding their place back.

Who else is on the selectors' radar? Alex Hales is a dominating opener in the shorter forms of the game who has recently scored runs in red ball cricket too. He deserves a chance but his first-class record is not nearly as good as Lyth's so there's no reason to think he'd do any better. Mark Stoneman has been mentioned but his record is not outstanding either.

Make do and mend
What about a makeshift opener? Moeen Ali or James Taylor. Or put Joe Root back there? No. Test matches are not the right place for jury-rigged team structures. Except perhaps for the series against Pakistan in the UAE where Moeen Ali could do a temporary job, England needs a high quality specialist opener with a good track record in first class cricket.

England needs Adam Lyth.

Anyway, it was instructive in 2013 that the hypothetical team I selected from both Ashes squads contained more Australians than Englishmen. After all, England had just won the Ashes against the worst Australian team in living memory. How could it be possible that man-for-man more Australians were deserving of a place in my team?

Subsequent events (5-0 to Australia in the 2013-14 tour) proved that I was in fact too kind to England in my selections.

Now it's 2015 and we could do the same exercise with another tour of South Africa in the offing. As I write this I have no idea who I'm going to pick so let's find out together:

First XI

1. Rogers
2. Warner
3. Smith *
4. Root
5. Moeen
6. Stokes
7. Nevill †
8. Siddle
9. Broad
10. Lyon
11. Anderson

Rest of the squad

• Cook
• Bairstow
• Voges
• Finn
• Wood
• M Marsh

Six Aussies in the first team

In 2013 there were 7 Australians in the combined side so this is an improvement for England but still shows that the result of the Ashes series favoured the home team a bit.

Cook makes an even better case for inclusion this time but still just loses out to Rogers and Warner (yes, I know Rogers has now retired but the point of this exercise is to pick the best players from the Ashes series).

Moeen Ali was wasted batting at 8. He would have scored more runs batting higher because he often got out flinging the bat when England were 9 wickets down. Even at 8 he still scored more runs than every other player except Cook and Root.

My team has a long tail with Peter Siddle coming in at 8. Neither middle order distinguished itself enough to give me a better option but Mitchell Marsh might have come in instad of Anderson who had a poor series except for Edgbaston.

First XI

1. Rogers
2. Warner
3. Bell
4. Pietersen
5. Clarke *
6. Watson
7. Haddin †
8. Swann
9. Siddle
10. Harris
11. Anderson

Rest of the squad

• Cook
• Root
• Bairstow
• Broad
• Bresnan
• Tremlett

Seven Aussies in the first team

Yep. I think Rogers is a marginal choice over Cook but it's my team so I'm picking him. All the subs are English, although again Bairstow was a marginal pick over Khawaja. Nobody making a cast iron case for this spot though.

No Prior?

Bairstow is the reserve keeper and in slightly better form with the bat.

Broad not in the first team?

No.

And Tremlett? He didn't even play.

The other candidates rules themselves out by playing. Onions was unlucky.

Part 1: Introduction <- You are here
Part 2: Technology
Part 3: Reviews & umpiring
Part 4: Proposals & conclusions

Aims & origin of the DRS

TV replays

Before the DRS was introduced we had two umpires on the field with nothing but their eyes and ears with which to make a decision. Television viewers, on the other hand, had the benefit of slow-motion replays and an increasing number of technology-led analysis tools that were developed to give the commentators something to talk about.

It was becoming clear that umpires' on-field decisions were not as good as had hitherto been believed. Pressure grew for umpires to have access to the same technological aids that the TV commentators were using.

Preventing the howler

Eventually a compromise was reached. Technology would be made available to the umpires on the basis that it would enable them to correct the occasional grossly bad decision. Human error is always a possibility, went the theory, and if we can cut out some of the outliers then the average decision quality will be improved.

The TV umpire was introduced, first to re-examine run outs and stumpings, then to check on bump-ball catches, finally to allow reviews of LBWs and caught behind decisions.

And somewhere along the way the original message was lost. The DRS was seen now not as a way of highlighting grossly bad decisions but as an oracle that could determine the exact truth of any decision. I don't know how this happened but it is now firmly stuck in people's minds that the DRS is supposed to be 100% accurate.

It demonstrably isn't of course. We'll go through some of its shortcomings below - there are many. But each time the DRS fails to give the viewer an unambiguous, incontrovertible report on what happened, the calls for its abolition grow louder. The pitchforks are brandished most vehemently of course by the supporters of whatever team was most recently "robbed" by the villainous DRS, but all the protests contribute to the noise and the background noise is growing.

Perception & complexity

Perhaps the message of the DRS is too complicated? The DRS can say "Yes, it's out", "No, you're probably right" or "I don't know either, frankly". Simple messages to me but this is a world where journalists want to know "Is that beefburger safe?" and a cabinet minister has to push one into his daughters' face to answer the question. Because it's not "yes" or "no" but "in all likelihood, yes, to the extent that the minor risk to my daughter is bearable".

Journalists and TV commentators seem unable to assume any numeracy at all in their audience. Although arts journalists are allowed to cite Goethe, Dvorak and Klimt without supporting explanations, any probability information that is presented to a viewer has to be reduced to a simple binary answer.

Here's Jagmohan Dalmiya of the BCCI[1] pretending to be stupid (he isn't):

We will accept DRS when technology is foolproof. There's nothing in between. Full stop. Let them come up with a system which is 100 per cent correct. They couldn't fix the Duckworth-Lewis problem in 15 years, what guarantee do we have about an error-free DRS? The Duckworth-Lewis method is beyond most of the players and administrators, let alone the common fans. I am still trying to figure out how can a team total be increased on basis of projection. The whole process is very complicated and confusing. And rather than solving the riddle, DRS creates more confusion in its present form. - Jagmohan Dalmiya doesn't see any hope for DRS in present form

There's a book to be written about the real reasons why the BCCI actually refuses to allow the DRS to be used in India's test matches[2], but concern for the fans is not one of them.

Perhaps that is simply a timid reaction to a new situation that the administrators fear the viewer will not understand. Perhaps the viewer is more intelligent than she is given credit for.

Here are some complex systems that seem to have survived despite the supposed ignorance of the sporting audience:

The Duckworth-Lewis method

The Duckworth-Lewis method is a way of adjusting the score of a limited overs cricket match that is affected by rain. Its aim is to ensure that the target that the team batting second is chasing is fair, taking into consideration all the factors that might have affected it.

The Duckworth-Lewis method is a rational answer to a complex problem. It could not be simpler without being unfair. It's outcome is a single number which is easy to understand.

It throws up some unusual results that are at first counter-intuitive (see Jagmohan Dalmiya's complaint above). These are easily explained by reference to the resources available to both teams, without any maths being needed.

The current Duckworth-Lewis par score can be calculated on a ball-by-ball basis and modern scoreboards incorporate this in their design. Both teams and the crowd know where they stand at all times.

Although the algorithm behind the Duckworth-Lewis method is opaque, it is now widely accepted by players and fans as a good approximation. Nobody is calling for it to be replaced by the previous crude adjustments based on average run-rate.

LBW law

The LBW law Law 36 (Leg before wicket) is complex and could in theory be simplified. Trying to explain the law to a cricket neophyte is troubling. "You mean it has to pitch in line with the stumps?". "No, it can pitch outside the off stump but not the leg stump". "But it has to hit the pads in line with the stumps, right?". "No, only if the batsman is playing a shot". Exit neophyte.

Calls for the LBW law to be reformed have existed throughout the history of cricket, and it has been reformed several times Wikipedia - Leg before wicket. But the law as it stands is widely respected despite its complexity and few, if any, are calling for change at the moment.

Offside rule in football

Cricket is not the only sport afflicted by irreducible complexity. Association Football's offside rule has always been a target for merriment based on people's poor understanding of it.

The offside rule has been reformed several times too Wikipedia - Offside (association football). But, while mocked, it has stood the test of time and will remain part of the game for the foreseeable future.

Too complicated?

So the evidence is that the DRS does not have to be presented as an infallible oracle. People (even "common fans") are willing to accept complexity if they know it is the minimum required for fairness. The DRS can produce and in some cases evaluate a body of evidence that can help a TV umpire assess a colleague's on-field decision. We can observe that process and appreciate the subtleties of what we are watching. It can be an interesting part of the game without patronising administrators and commentators telling us that we are too stupid to understand it.

[1] Board of Control for Cricket in India [2] As Deep Throat said: follow the money.

I'll publish this in sections - it's turned out to be a bit longer than I thought...

Contents

Part 1: Introduction
Part 2: Technology
Part 3: Reviews & umpiring
Part 4: Proposals & conclusions

To be done:

Hotspot

Evidence of absence

• Evidence of absence

Snicko

Sources of noise

Cum hoc ergo propter hoc

• Correlation does not imply causation

Real-time Snicko

Hawkeye

Margin of error & "umpire's call"

Outfield catches

Foreshortening

Run-outs & stumpings

Parallax

Umpire-led reviews

Reviews

Loss of reviews & "umpire's call"

• Professional players' view: James Anderson calls for end to DRS reviews lost to ‘umpire’s call’
• BCCI view: India in discussions over DRS

Umpire-led reviews

Current situation with run outs

Conflict of interest

Gaming the system - a strategy for selecting reviews

• Australia's poor choices
• Evidence of emerging strategies

Umpires' role & guidelines

Benefit of the doubt & "walking"

Authority of on-field umpire

Umpires' union

TV umpire's conflict of interest

Proposals

No loss of review for "umpire's call"

No umpire-led reviews

"Evidence of absence" principle

Communication & transparency

Conclusions

Evolution not revolution

Benefits for the paying customer

[1] Board of Control for Cricket in India [2] As Deep Throat said: follow the money.

I exaggerate. I'm not really going to tell you my password but I am going to tell you roughly how I created it. You may find it useful when thinking about your own passwords.

Over the years I got into the habit of using the same weakish password for every website I registered on. This was OK for a while then one of the sites I use got its password hashes published and within a few days they were cracked and my password was there for all to see.

I decided to do the sensible thing and create a strong, unique password for each site I use. The two problems I had are the same as everyone else:

1. I wanted the passwords to be easy for me to remember but hard to crack
2. I wanted them to be easy to enter on my phone

The second design goal was the one which made my approach slightly different to other ones you see around.

Here's how I make my passwords:

• To make a password that's unique for every site you need to incorporate some attribute of that site into the password
• If you have multiple accounts on the site, each should have a different password. So your algorithm needs to take account of your identity too.
• The password must be long enough to be strong but short enough to enter on a phone keyboard
• The password should only use the characters available unmodified on your phone keyboard (it's too painful to keep changing the keyset for me)

The characters I have available on my Android phone are the lowercase letters a-z and the punctuation keys ",", "." and space. Your phone may be different.

Here are the components of my password:

1. A word that's memorable to me, let's say "fart"
2. Some letters from the website's domain name
3. Some letters from my email address
4. Some mandatory punctuation

An example. Let's say I'm signing up to Facebook with the email address dominic@sayers.cc

My password would be fart fcbk ae.. ,,

It's made up as follows

1. My memorable word, "fart"
2. A space
3. The consonants from the web site's domain name, padded out to a length of 4 by full stops.
4. A space
5. The vowels from the domain name of my email address, padded out to a length of 4 by full stops
6. Some punctuation; a space followed by two commas.

So my password is 17 characters long: not perfect but strong enough for today's brute force attacks. It uses all the keyspace available without fiddling around on my phone's keyboard. It's unique for every account on every website (barring coincidences). It's easy to remember.

You can obviously vary these elements considerably to suit yourself. For instance you could use the last four letters in the website's name, maybe reversed or something. Rearrange the elements I've used.

This isn't my actual algorithm, by the way. Just a little bit like it.

Notes:

1. Please don't tell me about the XKCD cartoon. I know about it. I'm not going to type in a 28-character password on my phone.
2. A password using this algorithm doesn't lend itself to dictionary attacks or even rainbow table attacks. A brute force attack using today's hardware would take an inordinate amount of time according to this chart:

1. Alastair Cook*
2. Joe Root
3. Jonathan Trott
4. Ian Bell
5. James Taylor
6. Jonny Bairstow
7. Matt Prior†
8. Stuart Broad
9. Graeme Swann
10. James Anderson
11. Steven Finn

And the rest of the first-class squad:

1. James Tredwell
2. Graham Onions
3. Eoin Morgan
4. Tim Bresnan

Not 100% fit (mentally or physically) so not included: Ravi Bopara, Chris Tremlett

Unlucky: Monty Panesar, Chris Woakes, Samit Patel, Nick Compton

A bit boring and predictable for me, but I think that's a testament to the good stuff coming from Andy Flower and Geoff Miller.

Any new business or project is going to need an online presence. So you just need to go and register a domain with GoDaddy, right? Stop! First read this handy guide to how to get it right.

You need three basic things

1. A registered name
2. A way of telling the internet where your services are hosted
3. Hosting for your services (web site and email at least)

When you register a domain name, the registrar will want to do all these things for you. That way they get to keep all your money. But there are good reasons for keeping everything separate. Now read on...

Choosing a name

It's fun choosing your name, but it's not easy these days because all the good ones have already been taken. Here's some things to consider that you might not have thought of:

1. A .com domain name. D'oh!  Of course you thought of that. But if you didn't then you should. Plenty of people will try the .com version of your name even if you register it somewhere else.
2. A domain name in another top level domain. If you plan to allow users to make comments on your web site or even to upload files then consider having an alternate domain name that is not subject to US laws.
3. A Twitter handle. Twitter is pretty crowded these days and getting your name as a Twitter handle will not be easy. And Twitter doesn't have a secondary market in handles so you can't make someone else an offer for theirs..

Registering your name

You need to tell the gods of the internet that you have chosen your name. You do this through a registrar. Many companies will act as a registrar for you because it's money for old rope. They will also offer you other services like hosting and email. Here's why you shouldn't use the same company for everything.

Your domain names are your branding (and your trademark if you register it). It's your identity and you should keep tight control of it. If you were a big company then your domain names would be under the control of your legal or marketing department, not your techies. If you outsource your web development, for example, you don't want the developers owning your brand - what if you fall out with them?

So register the domain names yourself, but don't buy any other services from the registrar. Keep the credentials for your account at the registrar very safe. Nobody else needs to know them except you. Ever. If they say they do then they are trying to rip you off.

My recommended registrar: Gandi. Many countries have appalling internet legislation that compels internet service providers to pull your domain on the say-so of law enforcement agencies. France, where Gandi are domiciled, is no exception. But Gandi at least have a policy of pushing back when asked to do something that is not in the customer's interest. GoDaddy, on the other hand, are happy to hand over your data and your domain to the US security services at the request of a junior officer.

You'll also need to register a Twitter handle. To do this you'll need an email address, so let's defer this until we've got our domain email services working. Registering a Twitter handle is easy: you're just creating a Twitter account in the normal way.

Hosting your services

It's no good having a domain name unless you use it for something. Here's some of the services you might use it for:

1. Email You'll probably want email addresses at your domain (you@yourdomain.com). The easiest way to do this is Google Apps. To complete the setup of Google Apps you'll need to verify that the domain is yours - we'll cover this in the following section about the Domain Name System. When you've got your email accounts set up, don't forget to go and register your Twitter handle.
2. Website or web application Traditionally your website would be hosted by a provider that gave you access to a virtual machine. On this machine would be a web server and you could simply upload HTML or PHP pages to create your site. The problem with this way of doing things is that it's fiddly to keep track of what you uploaded and when. This model has been replaced somewhat by the idea of an application server. Providers like Heroku or AppFog will link to your web developer's source code control systems to create more effective release control system.
3. Blog Your blog is just another application, hosted by an application server. There are many companies that will do this for you. I happen to use Wordpress software hosted on my own old-fashioned virtual server. But Wordpress as a company will host your blog for you if you want.

Tying the bits together

As it stands we've got a domain name (good) and email addresses like phil@yourdomain.com.test-google-a.com (bleugh!) and maybe a website at http://yourdomain.herokuapp.com (yuk!)

This isn't what we want. We want emails like phil@yourdomain.com and our website to be at http://yourdomain.com - how do we do this?

The thing we need is the Domain Name System (DNS). Your domain has name servers. These servers tell the internet where your own services are located. It's a bit of internet plumbing that we need to use to tie it all together. Bear with me, it's not that hard.

I recommend using Cloudflare for your name servers. They are specialists at this and they do a free account that gives you all your basic needs, plus a lot of extra value on top. The Cloudflare account setup gives you all the help you need to get your services running on the right domain names.

When you've been through the Cloudflare setup you'll understand it a lot more. But the basic steps are:

1. Tell the DNS where your web hosts are
2. Tell the DNS where your mail servers are
3. Tell your registrar who is running your name servers

The last step is what achieves the important separation of responsibility between your registrar and your other service providers. It gives you the freedom to choose the best-of-breed hosting provider, email provider and DNS provider. You don't have to be tied to your registrar for these services any more.

If you're using Google Apps, the final step is to verify to Google that you own the domain. Google Apps setup will help you through this, but it's very easy with Cloudflare to set up a TXT record containing information provided by Google. You'll see what I mean when you try it.

Comments and suggestions appreciated.

I started using Flickr before they officially launched and I got my first Flickr Pro account in 2004 as a reward for testing their site. I'm still a customer today.

Dave Gorman has brought to light an issue that has existed without my knowing it since the US Digital Millennium Copyright Act came into force in 2009. Non-US users whose content is the subject of a DMCA takedown notice get their photos removed. For ever.

The most recent Internet Archive record of Dave Gorman's original page is here:

Sadly, this is what it now looks like.

Dave Gorman successfully appealed the takedown and Flickr graciously allowed him to repost his photo. But the URL has changed. All the inbound links are now broken and the comments now go back to March 2012 instead of January 2006.

Flickr doesn't work like this for US users. They get their photos restored as they were before (so Flickr can do this if they want to). They are choosing not to do so for me and all other non-US users, and as far as I can see they are not explaining why.

So goodbye Flickr after 8 years. I won't delete my account because that would break all the links to my photos, but I will stop being a customer. My new photos, and my money, will go elsewhere.

Dear Mr Fitzpatrick,

Given your comfortable majority I doubt you'll be too worried at the loss of one vote, but can I briefly give you my reasons why I, as a lifelong Labour voter, could not vote for you at the last election and will not be able to do so next time either?

I was never too worried about civil liberties until a few years ago. I'd actually be happy if there was a national identity card, for instance. I don't think I'm a paranoid libertarian nutter (fingers crossed). But I was dismayed by a number of measures introduced by the last Labour government starting in the climate of fear whipped up after the major terrorist incidents of 2001 and 2005.

Successive Labour Home Secretaries seemed to want to prove themselves more macho than their predecessor, and more tough on crime (rather than its causes) than the Opposition. Police were given more powers and used them to kettle protestors and prevent photography in public places. The UK acquiesced in the torture and long-term imprisonment without trial of British citizens. All this was not something that affected me personally so I noted it without allowing it to change my long-term loyalty to the party.

But then in the dying days of the government the Digital Economy Bill was introduced and passed without adequate debate. If there had been time to organize a coherent opposition to it then it would probably have suffered the same fate as the US SOPA and PIPA bills. But despite the help of people like Tom Watson, it was not possible to raise public awareness of the consequences of passing this bill until too late. It became the Digital Economy Act, the party lost the election (for unrelated reasons sadly) and we were stuck with it.

By contrast with the policing and security measures, the Act deals with something I know a little about: technology and the internet. I am a former technology director of an investment bank and am now the Chief Technology Officer of a financial services startup company.

My company's success depends on being able to respond in an agile way to the changing business environment. We can do this much faster than large corporates, however much they spend on R&D. We can certainly do this much faster than regulators and legislators, no matter much lobbyists spend on influencing them. But we are left standing by the bad guys, who can innovate before a legislator has even heard of the problem.

My contention is this:

• It's not possible to anticipate everything the bad guys might do, so you can't legislate for their future exploits.
• If you write over-broad legislation to counter this problem then you stifle legitimate business and prevent beneficial innovation.
• If you give broad powers to law enforcement agencies they will use them for purposes for which they were not originally intended. Sad but true.
• The law already makes crime criminal. More law will not actually change this.
• The central problem is a failure of established businesses to give the customer what they want nowadays.

The last point is key. Perhaps I could include some comparisons between the user's experience of piracy compared to a legitimate purchase:

Just to reiterate, I'm not supporting piracy. I'm saying if the content owners gave customers what they wanted, made it as easy as pirating it, and didn't try to retain an inappropriate level of control over what the customer did with the content then they might still have a business model.

I could go on; I'm quite angry about this. The Digital Economy Act will have no effect on piracy (or child pornography or anything else it's supposed to fix). I can think of ten ways of defeating its measures, so the bad guys can doubtless think of 100. Harriet Harman (and by implication your party) appears still to think otherwise: http://j.mp/zyvlFC

The Digital Economy Act reinforces the principal of a censored internet and adds disconnection for suspected infringers. What legislators should actually be doing is ensuring a free internet with access provided on a the same basis as other utilities.

I'd vote for a party that supported that.

Regards,

Dominic Sayers

Available domains

• refdata.org

• referen.org

• referen.se

Possible API structure

Let's take currency data as an example:

http://referen.org/currency/antigua.json

which gives you the details of Antigua's currency in JSON format. This would be the latest version of the format, and the current currency used in Antigua. For a specific data format you can add a version number, and for a specific data in history you can add a date:

http://referen.org/v1/2012/02/10/currency/antigua.json

As well as JSON, data would be available as XML or semantic HTML (the default), possibly as a microformat.

The API would try to be as smart as possible, like http://cricdata.org. So, you could specify antigua or USD or renminbi and it would try to give you an appropriate response. If ambiguous (e.g. dollar) then it would return a list of matching currencies.

Crowdsourcing

Some data, like the currency example, is available for free from an authoritative source (in this case http://www.currency-iso.org/dlisotable_a1.xml).

Other data is available from a reliable source such as http://opencorporates.com

But some data might need to be mastered here. In which case a wiki approach could be taken. In this case a reputation-based system would be good so that vandalism could be countered algorithmically. In other words, if you spot some bad data then you can flag it up (possibly anonymously, although an identified person's flagging would be taken more seriously than an anonymous one). Somebody whose contributions are flagged as bad will have their reputation compromised. Data contributed by somebody that is not flagged as bad will increase their reputation depending on how often it is eyeballed.

As a consumer of data you can choose to have all data, regardless of the reputation of the contributor. Or only high-reputation data. Or only authoritative data.

Data types

• Currencies
• Currency pairs
• Countries
• TLDs
• Second-level domains
• SIC codes

Obvious things like ISBNs, stock tickers, ISINs are actually fraught with licensing issues and generally poor design. But some attempt could be made.

Monetization

Dream on.

Times change.

I think it's time we seized control of something just as important in the 21st Century: the means of communication. Here's the story:

Dominic Sayers - The new Clause IV

Here's my simple guide to how internet domains work.

Name resolution

When you type www.example.com into a web browser, how does it know where to go to get the web page? As many people know, the answer is the DNS system. This translates www.example.com into the address of a web server. More mysterious stuff happens to connect your browser to that address, but we don't need to know about that today.

Your domain has a set of DNS records that say where on the internet your domain's resources (email servers, web servers etc.) actually live. You can change these records yourself, it's quite easy.

Resource records

Here are the DNS resource records you might need to change:

Address records (A): If your web server is at the internet address 192.0.43.10, for example, then you need to create an A record for www.example.com that points to that address.

Synonyms (CNAME): If there's a resource on a server that already has a name then you can simply tell the DNS that your resource is on the same server. This is how Google Apps creates a webmail site at mail.example.com; you simply create a CNAME record that points mail.example.com at ghs.google.com.. That's all you have to know.

Mail servers (MX): Your email is transferred by mail servers. These are the servers that actually move mail around (not to be confused with your webmail site). For Google Apps mail, the settings for the MX records are specified in Google Apps help pages.

Domain registration & name servers

Where are your DNS records kept? This is determined by the domain's registration setup.

When you first registered the domain you will have used a registrar like GoDaddy or Network Solutions (or Gandi, the registrar I recommend). For an annual fee this company will maintain your domain's existence on the internet and one other thing: determine where the domain's name servers are (the name servers hold the DNS records).

When you first registered the domain, the registrar will have used their own name servers for your domain. This is to make it easy for you, but also because they are control freaks.

This is important. Many registrars try to hide the fact that you don't have to use their name servers. In fact there are advantages in using a separate DNS provider from your registrar:

1. The domain registration should be managed by your organisation's legal or administrative department. This is the master key: in the event of a problem then the name servers can simply be changed to regain control of the domain.
2. Management of the DNS records can be delegated to a technical person, but it's really not difficult to do this yourself.
3. Specialist DNS service providers often add extra value: for instance Cloudflare will cache your web content and speed up web access for all your users.

Summary

So in a nutshell:

1. Your domain is registered with a registrar. Your account at the registrar controls where the domain's name servers are.
2. The name servers manage the DNS records for the domain. You should have a separate account with a DNS service provider to manage them.
3. Your web server will be controlled by an A record in the DNS setup.
4. Your webmail server will be controlled by a CNAME record in the DNS setup.
5. Your mail servers will be specified by MX records in the DNS setup.

If this isn't clear please leave a comment.