Internet Comments - Fred Hansen

Please place documents in this folder with written comments on what you think of the Internet system on ship. Please be honest, nice, and impress me with your good writing skills. --Matt

When talking of shipboard internet access, it is a challenge to be both honest and nice. Anyway, I'll give it a try.

An old nursery rhyme reads:

There was a little girl with a golden curl
Right in the middle of her forehead.
When she was good, she was very, very good.
And when she was bad, she was horrid.

The SAS version reads:

There was a little ship with an internet chip
Right in the middle of her foredeck
When she was good, she was very, very expensive
And when she was bad, a pain in the neck.

Overriding all other considerations about shipboard internet access is availability. It reeks. (That's honest, and about as nice as can be justified.) Service has a mean-time-between-failure of about two hours. (It fails four or five times every day. The ship's joiner is making us nice permanent reversible signs to indicate whether the net is up or down.) Each failure makes its impact:

whatever one was doing must be abandoned (frustrating!) or remembered (inducing tension!) and later restarted--thought threads are tangled
often the composition of a long email is interrupted--that work time is lost
if you try to login while the net is down, you face a long wait; you sit there frustrated, wondering if the system is down or just exhibiting one of its frequent annoying delays
more frustration ensues while checking to see if the internet is back yet
semester at sea loses just a bit more of whatever warm feelings users may once have had

Exacerbating the problem is the chunky nature of work time aboard ship. Work is interrupted by meals, classes, meetings, and shipboard activities. There is no good place to work, so interruptions also arise from loud conversations at any given time. When a work period is interrupted, resumption may be impossible until the next day. That the internet comes back in an hour or so is not usually useful.

From the random nature of the failures, it seems that the problem is software rather than hardware. Other software problems suggest an inadequate effort in that area:

Sometimes students are unable to disconnect.	Usage charges rise inexorably, like taxes. The system administrators spend hours at a time giving refunds.
It is not possible to change one's password.	Birthdays are published in the dean's memo. Ask a birthday girl how old she is and you can compute her password with no more effort than chewing cotton candy. (There is a change password button; it does not work)
The status box is necessary to disconnect. And yet it gets lost.	Without the status box, the damnable charge clock keeps ticking. (Also, the status box is a popup. Many uers have popups disabled, and yet the design gives no hint that they ust be enabled.)
It takes two separate clicks in two places to log out.	This is a good design except that it robs the user of a few more seconds every time one connects. It costs several cents just to stop getting charged.
New accounts must be created to convert students from free accounts to paid accounts.	Egads! Did someone judge this a rare event? 250 minutes over 100 days is two and a half minutes per day. When was the last time you, dear reader, got by with that little? Again the system administrators are saddled with extra work; work unnecessary in a properly designed system.
Several times a day the shipboard network runs out of IP addresses. No one can connect until manual intervention takes place.	More manual work for the IT administration. And they have to hear about the need from angry users. This note would have been posted sooner, but I could not get an address when I was ready.
On some mornings internet access is free.	Those in the know get free service, leaving the rest of us to foot more of the bill.
No attempt is made to connect via land lines while in port.	Why not? There would be savings for both passengers and ship's business.

{There are other instances of poorly designed systems aboard ship:
--Grade reporting requires the Registrar to manually fill in hundreds of bubbles on paper. The chance of error is huge; and students rarely learn of nor fix such errors.
--Field office systems are obfuscatory. See my report in public folders\FRED Hansen\FieldProgram.html.}

{While I've got you reading this, let me spend a minute on the wireless network aboard ship. It is inadequate; too many areas are without coverage. If there were even one networked area where one could work uninterrupted, it might be nearly tolerable, but there is not. It is essential to extend coveage to cabins, student and faculty alike. Routers cost less than a hundred dollars. The entire ship could be wired for a small fraction of what the food service spends on lettuce. (I suppose wireless is blocked by steel bulkheads. If so, we need another solution.)

{The network also lacks a choke point. The blue-noses aboard ship insist on turning off access during global studies; to do this Matt must personally visit each of the wireless routers.}

Internet service often boasts remarkable, even amazing, speed. This is good. Sometimes resource contention or other artifacts drop its speed to a crawl. This often happens near the end of a lengthy download, so one hangs a tantalizing ten percent away from completion. I'm personally accustomed to variable speed internet. It would not be too big a deal. BUT, I am paying for that long hiatus at the same rate I paid for high speed access. It is frustrating.

One of the remarkable features of the internet system aboard ship is that individuals have accounts. This is remarkable because the lack of accounts in other parts of the system is noticable and just plain wrong. Without accounts, there is no way to control who modifies which file on the public folders. (Only luck and well-behaved students have avoided malicious modifications. My own files have been modified by others at least twice.) Without student accounts, the only control we are able to wield is to avoid student modification of course files. Without accounts it is not possible to have shipboard email. One must search the ship or convince Kenn of the need to interrupt everyone aboard by blaring out a name on the intercom. And without accounts it is impossible to regulate or charge for use of scarce resources like paper and diskspace.

Since accounts are needed for the internet, they should be used more generally.

In the cold reaches of my heart there is a special place for a few of the positions listed on the Pittsburgh staff roster. I don't know the incumbents--they are probably great folks. They have probably personally complained long, hard, and futilely about the concerns I've expressed above. In any case, their titles suggest they might be useful to me and others aboard ship, and yet they seem not to be:

Information Systems Analyst	The systems needed more analysis.
Network and System Administrator	The ship's network is insufficient.
Help Desk Specialist	How do I get help from this person?

Finally, it is time to talk about the pricing of shipboard internet access. It is important to recognize that the internet has become a more valuable resource than a library. It is certainly more valuable--for most purposes, but not all--than the small library for which space is possible aboard ship. We should not consider abandoning our present library, but we must consider providing the internet resource as readily as we provide the library resource.

The high price charged for the internet has the effect of damping usage. It estranges personnel aboard ship from the wider academic community that has come to rely--even to live--on the internet. A lower price might well result in higher usage and thus higher total revenues.

The most egregious fault of the pricing system is that it charges for the wrong commodity. It is galling to process email and use few network resources while a neighbor is charged the same amount to download a file using many multiples of my own network bandwidth. If charging is to be done, it should be per packet or per byte, not just for connection time. By charging for connection time we are also lending the impression that that time is a quantity of value to be conserved. In fact, the connection is capable of a certain amount of traffic, independent of the number of users. Times when the network capacity is under-utilized are actually more expensive to the ship than times of full utilization.

(Of course, a full cost model must include both the capital cost of the equipment aboard ship and the communication costs imposed by he satellite manager. Again the charge-for-connection-time model is inadequate to reflect these charging realities.)

Conclusion

Shipboard internet access fails far too often. It imposes onerous mental and emotional costs upon users. Associated systems are ill-designed and poorly implemented. The costing model fails to recognize the academic value of the net and charges for the wrong quantities.

On a personal note, I have myself become dependent on world-wide communication and information access. The utter inadequacy of shipboard internet access was one of the major factors that caused me to seriously consider leaving the voyage prematurely.