# Posts by David

4 posts • joined 20 Apr 2008

### Taking a first bite out of Wolfram Alpha

#### Charge on an electron is rounded

Ask Wolfram Alpha what is the charge on an electron and you will get an answer of 1.6021765x10^-19 C.

But the best avaiable data is the CODATA 2006, which is what is reffered to by both NIST in the USA and NPL in the UK.

http://physics.nist.gov/cuu/Constants/codata.pdf

Wolfram Alpha have the value from CODATA 2006, but it has been rounded to 8 significant digits. The internationally agreed value is 1.602176487x10-19 C, not the value Wolfram Alpha gives.

So far from being a 'primary source' as Wolfram Research claim, it is taking primary data and rounding it. Wikipedia has the right value.

I'm a user of Mathematica and find that program impressive, although the free open-source Sage looks as though it will be a thorne in the side of Wolfram Reserach. I was impressed the day I first see Mathematica - more than 20 years ago. But I can't see much of use in Wolfram Alpha. It appears to be a poor man's Mathematica. For most things web relatved, google is far more useful.

I'm sure Wolfram Alpha will improve over time, but I can't ever see it being too succesful myself.

### Ballmer bitch slaps Vista

#### The quality of Vista is Microsoft's best defence against piracy

By making the software so bad, it reduces piracy. Perhaps this was Microsoft's plan. Seriously, I doubt there is the amount of piracy there would have been if the product was more useful.

In a moment of stupidity, I paid for an upgrade from Vista Business to Vista Ultimate on this £1620 laptop.

I've since fitted a new 300 GB disk, and now the machine runs Solaris x86 99% of the time. Since the new disk was fitted, I did put Vista back on it, but I can't be bothered to perform the 'upgrade' again to Ultimate.

If I do boot Vista, which is very rare, I might actually be breaking the law, as I have now upgraded to Ultimate and Business is not licensed. Somehow, I cant be too concerned.

Solaris x86 works well for me.

### Hitachi to go it alone on discs after all

#### MTBF - means something very different with disks.

In most fields, a MTBF of say 20,000 hours would mean the average life expectancy would be 20000/24=833 days, or 2 years and a couple of months.

Now look at data on hard drives and a MTBF of 1,000,000 hours is not uncommon. That is 157 years! But anyone who has handled 10 or more hard drives over a long period will know their average lifetime is nowhere near 157 years.

The 1,000,000 hours MTBF is only achieved if the drives are replaced (working or not) at the end of their service life, which is typically 5 years for a SCSI drive, and less for non-SCSI. So on average a SCSI drive will fail every 157 years, if you replace them in 5 years.

Most people do not realise the above. Check Seagate's web site. Other manufacturers will have similar methods.

For most users, a more meaningful rating would be the mean time before failure if you swithced them on and did not replace until they died. I tend to throw drives when either they stop working, or their capacity is so far below current capacities they are not worth having. I don't suppose I am too untypical. Used in this way, average lifetimes are certainly less than 10 years.

### Women love chocolate more than password security

#### Were the Business for Sofware Alliance (BSA) advising on the statistics?

It is not uncommon to see bad statistical data, but only the BSA could report stuff that is so statistically flawed. Remember the BSA saying:

"A 10 per cent reduction in the UK's software piracy rate would result in 34,000 new jobs, £11bn of economic growth and a £2.8bn increase in tax revenues"

This data about passwords is no more credible than what the BSA wrote.

A school child can see how flawed it is.