The Functional Database Before OLAP
Part 2 of 5The standalone-engine era of TM/1, 1983–1990.
This series examines TM/1 in its first form: a standalone, in-memory calculation engine sold for the IBM PC between 1983 and 1990, before OLAP had a name and before the acquisitions that eventually brought the product to IBM.
Reading time: 7–9 minutes
A Problem of Dates
The origin of TM/1 is genuinely contested, and the most honest thing a historian can do is refuse to give a single founding date.
What the sources actually support is a sequence of milestones, reported with varying confidence by witnesses of varying reliability.
The backstory is the most stable part.
By several accounts, Manny Perez conceived the idea while working at Exxon, where his department used an expensive, partly manual process to calculate supply and demand in order to set prices.
Frustrated, and — by his own cheerful description in a 2013 talk — “at heart a hacker,” he built something that integrated a database directly into a spreadsheet, so that stored data could be separated from the manipulation of it.
When VisiCalc appeared, he became convinced the spreadsheet was the right interface for the analytical engine he had in mind.
That much is consistent across the in-house history, the retrospective interviews, and the secondary accounts.
The dates are where the story fragments.
Five Sources, Five Beginnings
Lay the sources side by side and you get a five-year smear.
A vendor counter-history published by PARIS Technologies dates the conceptual start to early 1980 and offers a different etymology for the company name — Sinai plus Perez — than the one usually given for the product.
Wikipedia and several secondary histories describe a first version implemented on a mainframe timesharing system around 1981, before the PC version.
Cubewise’s Definitive History of TM1 — the closest thing to an authorised community account — places the formation of Sinper Corporation in early 1983, with TM/1 announced that summer at PC Expo in New York as the first Functional Database.
The project’s own version table dates commercial TM/1 Release 1.0 to 1984, with a list price of $795.
And a contemporaneous trade-press notice in InfoWorld, published on 28 January 1985, describes TM/1 as already available on the market at that same price.
Five sources.
Five different beginnings.
A Sequence, Not a Date
None of these dates is necessarily wrong.
The mistake is assuming they refer to the same event.
Conception, prototype, company formation, public announcement, and commercial shipment are different milestones.
A product can exist in one sense long before it exists in another.
The most defensible account is therefore not a date but a sequence:
- An idea takes shape around 1980–81.
- A prototype appears on a borrowed mainframe.
- Sinper is founded in early 1983.
- TM/1 is announced at PC Expo in summer 1983.
- Commercial Release 1.0 ships by 1984.
- The product is visibly on sale by January 1985.
This chronology accommodates all major sources without forcing them into artificial agreement.
For a historian, that is often preferable to selecting a single date and pretending the uncertainty does not exist.
The Name “TM/1”
The question of naming illustrates the same problem.
Cubewise reports that the product name TM/1 stood for Table Manager and was adopted shortly before PC Expo as the deadline approached.
PARIS Technologies offers a different story for Sinper, arguing that the company name itself was formed from the names Sinai and Perez.
These explanations are sometimes presented as competing narratives.
They are not.
One explains the company.
The other explains the product.
Both can be true simultaneously.
The historian’s task is not to force a choice where none is required.
How Sources Should Be Read
This is also a useful place to explain the method behind this series.
Not all sources answer the same questions equally well.
For questions of dating, contemporaneous evidence carries the greatest weight.
A trade-press notice from January 1985 reporting TM/1 as already available tells us more about when the product existed commercially than a recollection written thirty years later.
For questions of motive, however, retrospective accounts become valuable.
Why Manny Perez built the product.
Why he resisted certain terminology.
What problems he was trying to solve.
These are things only participants can explain.
Even partisan sources can be useful.
The PARIS Technologies history clearly advances its own interpretation of events, but it also provides names, claims, and leads that can be checked against other evidence.
Used carefully, such sources remain valuable.
The goal is not to find a perfect source.
The goal is to understand what each source can and cannot tell us.
A Historian’s Answer
So when was TM/1 founded?
The honest answer is that it depends on what exactly we mean by “founded.”
If we mean the idea, the answer may be around 1980.
If we mean the first prototype, perhaps 1981.
If we mean the company, early 1983.
If we mean the public announcement, summer 1983.
If we mean a commercially available product, then certainly by January 1985 and probably earlier.
The desire for a single founding date is understandable.
History, unfortunately, is often messier than that.
TM/1 did not appear all at once.
It emerged through a sequence of developments that stretched across several years.
Understanding that sequence matters because it reminds us that technologies are rarely born in a single moment.
They evolve.
And TM/1 was no exception.
Sources
Primary archive
- InfoWorld, 28 January 1985, New Analytical Tools Break The Mold (Doran Howitt)
- Internal TM1 history materials
- Project version table (TM1_Versions.xlsx)
- Multi-part Manny Perez interview
Public and secondary sources
- Cubewise, The Definitive History of TM1
- Wikipedia, IBM Planning Analytics
- PARIS Technologies, The Cancelled History of TM1
- Meet the Father of TM1 event recap (2013)
- TM1 Forum discussion thread
Series Navigation
← Mind the Gap: Why TM/1 Made Sense in the 1980s
Next →
A Cell-Oriented Heresy: What the Functional Model Actually Was