As per the discussion, one way of achieving this goal is to redefine the "shorthand" length dimension used throughout the standard document classes. Here's an extract of the relevant code snippets from ltxplain.dtx containing the abbreviated definition:
\newdimen\p@ \p@=1pt % this saves macro space and time
As such, issuing
\makeatletter\p@=1bp\makeatother% or \setlength{\p@}{1bp}
modifies the default 1pt reference to 1bp. Looking at article.cls (although other document classes are similar), many related lengths are set using \p@. Here's an excerpt:
\setlength\lineskip{1\p@}
\setlength\normallineskip{1\p@}
...
\setlength\parskip{0\p@ \@plus \p@}
...
\setlength\arraycolsep{5\p@}
\setlength\tabcolsep{6\p@}
\setlength\arrayrulewidth{.4\p@}
\setlength\doublerulesep{2\p@}
...
\setlength\fboxsep{3\p@}
\setlength\fboxrule{.4\p@}
...
\setlength\abovecaptionskip{10\p@}
\setlength\belowcaptionskip{0\p@}
...
\renewcommand\footnoterule{%
\kern-3\p@
\hrule\@width.4\columnwidth
\kern2.6\p@}
...
\setlength\columnsep{10\p@}
\setlength\columnseprule{0\p@}
including some macros like \maketitle and things associated with indexing. So, issue the size change before \documentclass in order to let the effect filter through. You would still "miss" some \p@-related definitions though, as may be seen by viewing latex.ltx.
As a quick way to check the difference in the default pt and modified bp measurements (in lmodern) is using printlen. Here's a brief example with focus on the character X:

\documentclass[12pt]{article}
\usepackage{lmodern}% http://ctan.org/pkg/lmodern
\usepackage{printlen}% http://ctan.org/pkg/printlen
\begin{document}
\uselengthunit{pt} \renewcommand{\arraystretch}{1.5}%
\setbox0=\hbox{\fontsize{12pt}{14pt}\selectfont X}% pt measurement
\setbox1=\hbox{\fontsize{12bp}{14pt}\selectfont X}% bp measurement
\begin{tabular}{ccc}
X & width & height \\ \hline
\verb!pt! & \printlength{\wd0} & \printlength{\ht0} \\
\verb!bp! & \printlength{\wd1} & \printlength{\ht1} \\ \hline
\end{tabular}
\end{document}
The difference in width is around 0.04pt and 0.03pt in height, which translates to about 0.01mm - a roughly 0.3% increase (~ 72.27/72-1). This is virtually negligible to the naked eye at regular font sizes.
Paragraph construction is altered using 12bp rather than 12pt, and therefore also hyphenation. Here's an example showing the effect:

\documentclass[12pt]{article}
\usepackage[margin=0.5in]{geometry}% http://ctan.org/pkg/geometry
\usepackage{lmodern}% http://ctan.org/pkg/lmodern
\usepackage{lipsum}% http://ctan.org/pkg/lipsum
\begin{document}
\renewcommand{\arraystretch}{1.5}%
\begin{tabular}{p{0.4\linewidth}p{0.4\linewidth}}
\verb!12pt! font & \verb!12bp! font \\ \hline
%\fontsize{12pt}{14pt}\selectfont% pt measurement
\lipsum[1] &
\fontsize{12bp}{14pt}\selectfont% bp measurement
\lipsum[1]
\end{tabular}
\end{document}
;)– doncherry Nov 07 '11 at 20:42\documentclass{article} \usepackage{lmodern} \usepackage[T1]{fontenc} \begin{document} \fontsize{12bp}{14bp}\selectfont foo \fontsize{12pt}{14pt}\selectfont foo \end{document}produces a document with 2 different font sizes forfoo: the first is12bpand the second is12pt.lmodernhas this modification by default. Is this what you're after? – Werner Nov 07 '11 at 20:55bp. What I actually had in mind (i.e. how I solved it) is telling LaTeX to always use Word's point instead of its own. – doncherry Nov 07 '11 at 21:06bpinstead ofpt. But I don't think the difference can be appreciated by the naked eye. An "A" at 12pt is 8.13588pt high, while it is 8.1664pt at 12bp. The difference is 0.01 millimeters. – egreg Nov 07 '11 at 21:19\makeatletter \p@=1bp \makeatotherin the document preamble? – Werner Nov 07 '11 at 21:23\setlength{\p@}{1bp}. It does end up in a few different hyphenations, but it'd be a coincidence if they added up to an extra line or the like. – doncherry Nov 07 '11 at 21:26\p@. What I was saying is that it's almost impossible that somebody can spot the difference between 12pt and 12bp size. Particularly if they require MS Word documents. :) – egreg Nov 07 '11 at 21:32\p@but chosen relative to12ptthat would "get out of relation" (on a negligible scale as well) if we redefined\p@only? – doncherry Nov 07 '11 at 21:44\p@and not ofpt. But one never knows. Remember, on the contrary, that 17 point size is, for LaTeX, 17.28. I wouldn't worry at all. – egreg Nov 07 '11 at 21:50size12.clo, right? If a dimension isn't altered because of the12ptoption, it won't need altering if we usebpinstead ofpt. I looked into that file but I don't understand all of it. – doncherry Nov 07 '11 at 22:01\p@=1bprelies on anything other than the document class to use\p@rather thanpt. Evenltplainspecifies the use of\p@as an "abbreviation" for1pt(or more specifically, "this saves macro space and time"). Perhaps it's enough to assume that the cases where some other package usesptinstead ofp@is negligible. – Werner Nov 07 '11 at 22:16ptdirectly? Then I guess if one of you guys wants to, you could write up an answer with our gathered findings. – doncherry Nov 07 '11 at 22:21TeXhas a command\magnification, which was disabled inLaTeX. The magnification produced this way is not really a 'typographically correct' resizing of the font, but for such subtle differences, it should do a perfect job. Other than that, I think nobody would guess it is not a12bpfont, unless an actual12bpfont is placed next to it. (This may happen in documents where some pages are typeset withLaTeXand others in MS Word... but then again, probably other features/elemets will differ much more spectacularly.) :) – Count Zero Nov 08 '11 at 00:23