Some time ago (November 2002) I had a few beers one night and decided to
try to guess the *future* prices of Sun 450MHz X1195A CPUs (used in
Ultra 60s, 80s and others).

Some may recall the post, as it got a few comments. I based it on what
I'd paid for a CPU in mid 2000, what was the going rate in November
2002, and fitted an exponential. The exponential was used for no other
reason than it seemed the most likely - I'd not attempted to verify it.

http://groups.google.co.uk/group/com...9927ddf4fcd4eb

I came up with these prices:

1 month - Dec 2002 = $415
2 months - Jan 2003 = $383
3 months - Feb 2003 = $353
4 months - Mar 2003 = $326
5 months - Apr 2003 = $301
6 months - May 2003 = $278
9 months - Aug 2003 = $219
12 months - Nov 2003 = $172
18 months - May 2004 = $107
24 months - Nov 2004 = $67
5 years - Nov 207 = $3.70
10 years - Nov 2012 = 3 cents

Three or four people have commented to me this has been reasonably
accurate.

Today, I decided to check what X1195A's are selling on eBay. As always,
there is a spread of prices. The lowest price one on eBay for June 2005
is this one

http://cgi.ebay.com/ws/eBayISAPI.dll...3747 777&rd=1

So my estimate, made 31 months ago, which would suggest $37.68 was
pretty damm good, as this one sold at $39 !! An error of only 32 cents!!
Typically they are selling a bit higher at around $48, so my estimate
is about 30% low.

This is not bad bad considering the prices have dropped by more than 900%.

Comments ? - Other than I am a complete nutter !!

Since the original data did not give monthly values for more than 12
months ahead, but only at, here is a complete list. The same equation as
I used before

price = 450 exp(- 0.08*months) is still used.

I suspect at some point prices will rise. In 1000 years time, an X1195A
will probably be a sought after antique.

/* Historical estimates - future prices, which were computed in November
2002. */
0 months - Nov 2002 $450.00
1 months - Dec 2002 $415.40
2 months - Jan 2003 $383.46
3 months - Feb 2003 $353.98
4 months - Mar 2003 $326.77
5 months - Apr 2003 $301.64
6 months - May 2003 $278.45
7 months - Jun 2003 $257.04
8 months - Jul 2003 $237.28
9 months - Aug 2003 $219.04
10 months - Sept 2003 $202.20
11 months - Oct 2003 $186.65
12 months - Nov 2003 $172.30
13 months - Dec 2003 $159.05
14 months - Jan 2004 $146.83
15 months - Feb 2004 $135.54
16 months - Mar 2004 $125.12
17 months - Apr 2004 $115.50
18 months - May 2004 $106.62
19 months - Jun 2004 $98.42
20 months - Jul 2004 $90.85
21 months - Aug 2004 $83.87
22 months - Sept 2004 $77.42
23 months - Oct 2004 $71.47
24 months - Nov 2004 $65.97
25 months - Dec 2004 $60.90
26 months - Jan 2005 $56.22
27 months - Feb 2005 $51.90
28 months - Mar 2005 $47.91
29 months - Apr 2005 $44.22
30 months - May 2005 $40.82

/* Future pedicted prices */
31 months - Jun 2005 $37.68
32 months - Jul 2005 $34.79
33 months - Aug 2005 $32.11
34 months - Sept 2005 $29.64
35 months - Oct 2005 $27.36
36 months - Nov 2005 $25.26
37 months - Dec 2005 $23.32
38 months - Jan 2006 $21.53
39 months - Feb 2006 $19.87
40 months - Mar 2006 $18.34
41 months - Apr 2006 $16.93
42 months - May 2006 $15.63
43 months - Jun 2006 $14.43
44 months - Jul 2006 $13.32
45 months - Aug 2006 $12.30
46 months - Sept 2006 $11.35
47 months - Oct 2006 $10.48
48 months - Nov 2006 $9.67
49 months - Dec 2006 $8.93
50 months - Jan 2007 $8.24
51 months - Feb 2007 $7.61
52 months - Mar 2007 $7.02
53 months - Apr 2007 $6.48
54 months - May 2007 $5.98
55 months - Jun 2007 $5.52
56 months - Jul 2007 $5.10
57 months - Aug 2007 $4.71
58 months - Sept 2007 $4.35
59 months - Oct 2007 $4.01
60 months - Nov 2007 $3.70
61 months - Dec 2007 $3.42
62 months - Jan 2008 $3.16
63 months - Feb 2008 $2.91
64 months - Mar 2008 $2.69
65 months - Apr 2008 $2.48
66 months - May 2008 $2.29
67 months - Jun 2008 $2.12
68 months - Jul 2008 $1.95
69 months - Aug 2008 $1.80
70 months - Sept 2008 $1.66
71 months - Oct 2008 $1.54
72 months - Nov 2008 $1.42
73 months - Dec 2008 $1.31
74 months - Jan 2009 $1.21
75 months - Feb 2009 $1.12
76 months - Mar 2009 $1.03
77 months - Apr 2009 $0.95
78 months - May 2009 $0.88
79 months - Jun 2009 $0.81
80 months - Jul 2009 $0.75
81 months - Aug 2009 $0.69
82 months - Sept 2009 $0.64
83 months - Oct 2009 $0.59
84 months - Nov 2009 $0.54
85 months - Dec 2009 $0.50
86 months - Jan 2010 $0.46
87 months - Feb 2010 $0.43
88 months - Mar 2010 $0.39
89 months - Apr 2010 $0.36
90 months - May 2010 $0.34
91 months - Jun 2010 $0.31
92 months - Jul 2010 $0.29
93 months - Aug 2010 $0.26
94 months - Sept 2010 $0.24
95 months - Oct 2010 $0.23
96 months - Nov 2010 $0.21
97 months - Dec 2010 $0.19
98 months - Jan 2011 $0.18
99 months - Feb 2011 $0.16
100 months - Mar 2011 $0.15
101 months - Apr 2011 $0.14
102 months - May 2011 $0.13
103 months - Jun 2011 $0.12
104 months - Jul 2011 $0.11
105 months - Aug 2011 $0.10
106 months - Sept 2011 $0.09
107 months - Oct 2011 $0.09
108 months - Nov 2011 $0.08
109 months - Dec 2011 $0.07
110 months - Jan 2012 $0.07
111 months - Feb 2012 $0.06
112 months - Mar 2012 $0.06
113 months - Apr 2012 $0.05
114 months - May 2012 $0.05
115 months - Jun 2012 $0.05
116 months - Jul 2012 $0.04
117 months - Aug 2012 $0.04
118 months - Sept 2012 $0.04
119 months - Oct 2012 $0.03
120 months - Nov 2012 $0.03

Thinking about it more, there has got to be some low value below which
an item will never fall. Nobody is going to bother selling a CPU at
$0.03. It is simply not worth their time.

Extending this sort of thing to other computer hardware, could make an
interesting project for someone.

Dave <nospam@nowhere.com> wrote:
> This is not bad bad considering the prices have dropped by more than 900%.

> Comments ? - Other than I am a complete nutter !!

Is it even possible for something to drop by 900%?-) Seems that a drop
of 100% would leave the thing at $0.00. So, perhaps 99.somenumberof9s%?

rick jones

Rick Jones wrote:
> Dave <nospam@nowhere.com> wrote:
>
>>This is not bad bad considering the prices have dropped by more than 900%.
>
>
>>Comments ? - Other than I am a complete nutter !!
>
>
> Is it even possible for something to drop by 900%?-) Seems that a drop
> of 100% would leave the thing at $0.00. So, perhaps 99.somenumberof9s%?
>
> rick jones

Sorry, they have lost 90% of their value, not 900.

Actually, I just spotted another error too. I said the one that sold at
$39 was within 32 cents of my projected price of $37.68. It was of
course $1.32.

My predictions are if anything a bit under the selling prices (and it
seems to have stayed that way for any examples I've seen over the last 2
and a half years). But the error is pretty small considering I did this
in Nov 2002, based on the average price in Nov 2002 and the price I'd
paid for one in mid 2001.

Dave wrote:

> Some time ago (November 2002) I had a few beers one night and decided to
> try to guess the *future* prices of Sun 450MHz X1195A CPUs (used in
> Ultra 60s, 80s and others).
>
> Some may recall the post, as it got a few comments. I based it on what
> I'd paid for a CPU in mid 2000, what was the going rate in November
> 2002, and fitted an exponential. The exponential was used for no other
> reason than it seemed the most likely - I'd not attempted to verify it.
>
> http://groups.google.co.uk/group/com...9927ddf4fcd4eb
>
>
> I came up with these prices:
>
> 1 month - Dec 2002 = $415
> 2 months - Jan 2003 = $383
> 3 months - Feb 2003 = $353
> 4 months - Mar 2003 = $326
> 5 months - Apr 2003 = $301
> 6 months - May 2003 = $278
> 9 months - Aug 2003 = $219
> 12 months - Nov 2003 = $172
> 18 months - May 2004 = $107
> 24 months - Nov 2004 = $67
> 5 years - Nov 207 = $3.70
> 10 years - Nov 2012 = 3 cents
>
> Three or four people have commented to me this has been reasonably
> accurate.
>
> Today, I decided to check what X1195A's are selling on eBay. As always,
> there is a spread of prices. The lowest price one on eBay for June 2005
> is this one
>
> http://cgi.ebay.com/ws/eBayISAPI.dll...3747 777&rd=1
>
>
> So my estimate, made 31 months ago, which would suggest $37.68 was
> pretty damm good, as this one sold at $39 !! An error of only 32 cents!!
> Typically they are selling a bit higher at around $48, so my estimate
> is about 30% low.
>
> This is not bad bad considering the prices have dropped by more than 900%.
>
> Comments ? - Other than I am a complete nutter !!
>
> Since the original data did not give monthly values for more than 12
> months ahead, but only at, here is a complete list. The same equation as
> I used before
>
> price = 450 exp(- 0.08*months) is still used.
>
> I suspect at some point prices will rise. In 1000 years time, an X1195A
> will probably be a sought after antique.
>
> /* Historical estimates - future prices, which were computed in November
> 2002. */
> 0 months - Nov 2002 $450.00
> 1 months - Dec 2002 $415.40
> 2 months - Jan 2003 $383.46
> 3 months - Feb 2003 $353.98
> 4 months - Mar 2003 $326.77
> 5 months - Apr 2003 $301.64
> 6 months - May 2003 $278.45
> 7 months - Jun 2003 $257.04
> 8 months - Jul 2003 $237.28
> 9 months - Aug 2003 $219.04
> 10 months - Sept 2003 $202.20
> 11 months - Oct 2003 $186.65
> 12 months - Nov 2003 $172.30
> 13 months - Dec 2003 $159.05
> 14 months - Jan 2004 $146.83
> 15 months - Feb 2004 $135.54
> 16 months - Mar 2004 $125.12
> 17 months - Apr 2004 $115.50
> 18 months - May 2004 $106.62
> 19 months - Jun 2004 $98.42
> 20 months - Jul 2004 $90.85
> 21 months - Aug 2004 $83.87
> 22 months - Sept 2004 $77.42
> 23 months - Oct 2004 $71.47
> 24 months - Nov 2004 $65.97
> 25 months - Dec 2004 $60.90
> 26 months - Jan 2005 $56.22
> 27 months - Feb 2005 $51.90
> 28 months - Mar 2005 $47.91
> 29 months - Apr 2005 $44.22
> 30 months - May 2005 $40.82
>
> /* Future pedicted prices */
> 31 months - Jun 2005 $37.68
> 32 months - Jul 2005 $34.79
> 33 months - Aug 2005 $32.11
> 34 months - Sept 2005 $29.64
> 35 months - Oct 2005 $27.36
> 36 months - Nov 2005 $25.26
> 37 months - Dec 2005 $23.32
> 38 months - Jan 2006 $21.53
> 39 months - Feb 2006 $19.87
> 40 months - Mar 2006 $18.34
> 41 months - Apr 2006 $16.93
> 42 months - May 2006 $15.63
> 43 months - Jun 2006 $14.43
> 44 months - Jul 2006 $13.32
> 45 months - Aug 2006 $12.30
> 46 months - Sept 2006 $11.35
> 47 months - Oct 2006 $10.48
> 48 months - Nov 2006 $9.67
> 49 months - Dec 2006 $8.93
> 50 months - Jan 2007 $8.24
> 51 months - Feb 2007 $7.61
> 52 months - Mar 2007 $7.02
> 53 months - Apr 2007 $6.48
> 54 months - May 2007 $5.98
> 55 months - Jun 2007 $5.52
> 56 months - Jul 2007 $5.10
> 57 months - Aug 2007 $4.71
> 58 months - Sept 2007 $4.35
> 59 months - Oct 2007 $4.01
> 60 months - Nov 2007 $3.70
> 61 months - Dec 2007 $3.42
> 62 months - Jan 2008 $3.16
> 63 months - Feb 2008 $2.91
> 64 months - Mar 2008 $2.69
> 65 months - Apr 2008 $2.48
> 66 months - May 2008 $2.29
> 67 months - Jun 2008 $2.12
> 68 months - Jul 2008 $1.95
> 69 months - Aug 2008 $1.80
> 70 months - Sept 2008 $1.66
> 71 months - Oct 2008 $1.54
> 72 months - Nov 2008 $1.42
> 73 months - Dec 2008 $1.31
> 74 months - Jan 2009 $1.21
> 75 months - Feb 2009 $1.12
> 76 months - Mar 2009 $1.03
> 77 months - Apr 2009 $0.95
> 78 months - May 2009 $0.88
> 79 months - Jun 2009 $0.81
> 80 months - Jul 2009 $0.75
> 81 months - Aug 2009 $0.69
> 82 months - Sept 2009 $0.64
> 83 months - Oct 2009 $0.59
> 84 months - Nov 2009 $0.54
> 85 months - Dec 2009 $0.50
> 86 months - Jan 2010 $0.46
> 87 months - Feb 2010 $0.43
> 88 months - Mar 2010 $0.39
> 89 months - Apr 2010 $0.36
> 90 months - May 2010 $0.34
> 91 months - Jun 2010 $0.31
> 92 months - Jul 2010 $0.29
> 93 months - Aug 2010 $0.26
> 94 months - Sept 2010 $0.24
> 95 months - Oct 2010 $0.23
> 96 months - Nov 2010 $0.21
> 97 months - Dec 2010 $0.19
> 98 months - Jan 2011 $0.18
> 99 months - Feb 2011 $0.16
> 100 months - Mar 2011 $0.15
> 101 months - Apr 2011 $0.14
> 102 months - May 2011 $0.13
> 103 months - Jun 2011 $0.12
> 104 months - Jul 2011 $0.11
> 105 months - Aug 2011 $0.10
> 106 months - Sept 2011 $0.09
> 107 months - Oct 2011 $0.09
> 108 months - Nov 2011 $0.08
> 109 months - Dec 2011 $0.07
> 110 months - Jan 2012 $0.07
> 111 months - Feb 2012 $0.06
> 112 months - Mar 2012 $0.06
> 113 months - Apr 2012 $0.05
> 114 months - May 2012 $0.05
> 115 months - Jun 2012 $0.05
> 116 months - Jul 2012 $0.04
> 117 months - Aug 2012 $0.04
> 118 months - Sept 2012 $0.04
> 119 months - Oct 2012 $0.03
> 120 months - Nov 2012 $0.03

Using eBay prices to measure the accuracy of your predictions is
reasonable, but prices on local buy'n'sell newsgroups tend to be
considerably lower, and may well be the primary market for lower priced
items - particularly when listing and shipping costs approach or exceed
sale prices.

For some time my son has fully covered all costs associated with his
mountain biking hobby by collecting discarded PCs from the kerb and
parting them out through local newsgroups. He frequently realises $30+
on a PC the original owner tossed out as having no value. The PC parts
almost always sell within a day or two.

OTOH, I have a basement full of older Sun gear that sells very slowly
regardless of how I price it. Exponential price decay sounds reasonable
on the surface, but does not correlate with my experience in the market.

Sunny

"Dave" <nospam@nowhere.com> wrote...
> Some time ago (November 2002) I had a few beers one night and decided to
> try to guess the *future* prices of Sun 450MHz X1195A CPUs (used in
> Ultra 60s, 80s and others).
>
<snipped the start of an interesting thesis on Sun prices>
Was this all brought about by thinking about the price of
SBus video capture boards?;-)

I mean...
Who wants to know about old age in computers!;-)
It's all the latest, fastest, cheapest, smallest...
It's all the sexy, young things - right?

Generally - these days - the decline in prices is something like
you show.
But, say, 25-30 years ago the decline was much more gentle.
Admittedly, then, the second-user life of the kit might have
been another 10 to 15 years, on top of the 10 years or so the
original user got from the kit. You could buy kit - then -
6 months or more prior to it being de-installed: you knew the
price wouldn't change that much and you knew there would
still be a market for it. Sometimes, even, the manufacturer
increased the new price and the kit - when it became available
- was worth a lot more than you bought it for.
(Oh, happy days!;-)
Now, it's risky to buy kit as little as 2 weeks in advance!

But you don't allow for an upward curve towards the end of the
life - before the price (and the part) disappears into oblivion.
That - for us - is the more interesting (and profitable) area
related to holding stocks of old parts.

This upward movement is getting less pronounced, as the
life-cycle of the computer kit gets shorter, but it is still there.
So, I don't think you could give a simple, fixed equation for the
decline in the price. It would have to be related to the life of
the product, the volumes shipped, geographic cover, etc, etc.
But I think it should be possible to arrive at something useful...

The scenario for the price increase is that the manufacturer
no longer stocks the part (or maybe no-one there even knows
about it); less and less people (anywhere in the world) are
liable to have the part; so it gets more difficult to find; the
possibility of repair also disappears (as the people who knew
about it and the equipment to repair it goes out of use); the
user needs to keep the kit running; replacing the kit means
costs beyond simply the cost of the new hardware (replacing
software, re-wiring and re-fitting, etc).

So the price declines until it hits a 'floor'. It then bobbles
along for a while. And then starts rising again. The 'floor' -
for us - would be about $50. For others it may be a lot less.
But I can't see anyone selling individual, tested parts for $10
or $15. (Especially if it were an uncommon part, say, a video
capture, like a Sun 501-2253!;-)

People seem surprised sometimes that the price of an old
part has gone *up*!

'No, C'mon! I bought this part 2 years ago for only $100.'
'But... But... It's the last one left... In the whole world!'
'But my power-station... She will not work!'
'And how much will that cost?'
'It'll destroy an area of 4 miles in diameter! And kill 1000s!'
'And is that cheaper than buying the part?'

--
Tom Tobin.
Black humour creativity warning!
The last paragraph is not now, and never has been, a reality.

On Thu, 23 Jun 2005 01:03:43 +0100,
Dave <nospam@nowhere.com>, in
<42b9fc5f@212.67.96.135> wrote:
>+ Thinking about it more, there has got to be some low value below which
>+ an item will never fall.

Correct. At most, it'll approach and oscillate around some asymptotic
limit. Prices may in fact creep back up a bit, once the manufacturer
ceases production if there's adequate demand.

That limit could be estimated as A + B + C:

A = $(fixed cost of production per unit)
B = $(minimum manufacturer profit margin)
C = $(minimum reseller profit margin)

James
--
Consulting Minister for Consultants, DNRC
I can please only one person per day. Today is not your day. Tomorrow
isn't looking good, either.
I am BOFH. Resistance is futile. Your network will be assimilated.

Tom Tobin wrote:
> "Dave" <nospam@nowhere.com> wrote...
>
>>Some time ago (November 2002) I had a few beers one night and decided to
>>try to guess the *future* prices of Sun 450MHz X1195A CPUs (used in
>>Ultra 60s, 80s and others).
>>
>
> <snipped the start of an interesting thesis on Sun prices>
> Was this all brought about by thinking about the price of
> SBus video capture boards?;-)

I don't think so, although what bought me to bring it up again I do not
know.

I did think the person was asking a bit much for a card to be shipped
for a tenna including carriage.

> Generally - these days - the decline in prices is something like
> you show.
> But, say, 25-30 years ago the decline was much more gentle.

But was it exponential? I determined the rate of decline (the k=0.08)
based on two points. But if you decrease that, the the decay is still
exponential, but just slower.

> Now, it's risky to buy kit as little as 2 weeks in advance!

Yes, I can see that.

> But you don't allow for an upward curve towards the end of the
> life - before the price (and the part) disappears into oblivion.
> That - for us - is the more interesting (and profitable) area
> related to holding stocks of old parts.

Another aspect too is that the price might well become negative, where
you actually have to pay a company to clear the kit out.

> This upward movement is getting less pronounced, as the
> life-cycle of the computer kit gets shorter, but it is still there.
> So, I don't think you could give a simple, fixed equation for the
> decline in the price. It would have to be related to the life of
> the product, the volumes shipped, geographic cover, etc, etc.
> But I think it should be possible to arrive at something useful...

I'm sure there could be some intereting work for someone to do in as an
academic project.

> The scenario for the price increase is that the manufacturer
> no longer stocks the part (or maybe no-one there even knows
> about it); less and less people (anywhere in the world) are
> liable to have the part; so it gets more difficult to find; the
> possibility of repair also disappears (as the people who knew
> about it and the equipment to repair it goes out of use); the
> user needs to keep the kit running; replacing the kit means
> costs beyond simply the cost of the new hardware (replacing
> software, re-wiring and re-fitting, etc).

Yes, I can see prices can increase. As I expect it clould well do with
that Sun video card someone is looking for, as I imagine that was a rare
beast.

> So the price declines until it hits a 'floor'. It then bobbles
> along for a while. And then starts rising again. The 'floor' -
> for us - would be about $50.

> For others it may be a lot less.
> But I can't see anyone selling individual, tested parts for $10
> or $15. (Especially if it were an uncommon part, say, a video
> capture, like a Sun 501-2253!;-)

I'd not read down this to hit when I made my comments above! But I
suspect he is chancing his luck trying to get them postage paid for 10
pounds.

> People seem surprised sometimes that the price of an old
> part has gone *up*!

Let's hope SPARC 20's do - I have 5 of them !!

> 'No, C'mon! I bought this part 2 years ago for only $100.'
> 'But... But... It's the last one left... In the whole world!'
> 'But my power-station... She will not work!'
> 'And how much will that cost?'
> 'It'll destroy an area of 4 miles in diameter! And kill 1000s!'
> 'And is that cheaper than buying the part?'

It must be nice to be in that position sometimes.

> --
> Tom Tobin.
> Black humour creativity warning!
> The last paragraph is not now, and never has been, a reality.
>
>

Sunny wrote:

> Using eBay prices to measure the accuracy of your predictions is
> reasonable, but prices on local buy'n'sell newsgroups tend to be
> considerably lower, and may well be the primary market for lower priced
> items - particularly when listing and shipping costs approach or exceed
> sale prices.

Using eBay was really the only way to do it, since I based the original
equation on eBay prices. Had I based that on newsgroup prices, then

> For some time my son has fully covered all costs associated with his
> mountain biking hobby by collecting discarded PCs from the kerb and
> parting them out through local newsgroups. He frequently realises $30+
> on a PC the original owner tossed out as having no value. The PC parts
> almost always sell within a day or two.

It's good to see someone doing that, but whilst I don't sell stuff for a
living, I do have a lower limit which is determined by the effort to
send it. Should someone want an sbus card that I can stick in a jiffy
bag and send at their risk, I would probably sell it for 5 UKP ($10).
But if its a computer I have to box up and take to the post office, then
its a very different matter altogether.

> OTOH, I have a basement full of older Sun gear that sells very slowly
> regardless of how I price it. Exponential price decay sounds reasonable
> on the surface, but does not correlate with my experience in the market.