Good to know. How do non-translators see Translation Machines.

When deciding whether or not to use machine translation, consider the value of your content. If it’s highly valuable, mission-critical content, such as marketing messaging, MT will not accomplish your objectives. However, if it’s something of lower value like user support material or internal-facing content, MT with human post-editing may be the solution for you.
Alyssa Paris, marketing manager with Acclaro, a global translating business, offered the following tips for successfully using machine translation:
1. Make sure your source text is cleanly written and correct, free of idiomatic phrasings and generally machine translation-friendly. It’s also best if there’s a uniform tone throughout.
2. Your language vendor should establish upfront a dictionary of key terms. They should then have it reviewed and confirmed by all language stakeholders at your company.
3. Determine which approach — rules-based, statistical or hybrid — is best for your project. Statistical engines are appropriate if you have huge volumes of existing bilingual content. Otherwise, consider rules-based or a hybrid solution.
4. Invest in human post-editing for improved accuracy and quality. However, the editors should only change what is lexically essential to ensure understanding.
5. Have realistic expectations for the quality of the content you’re going to achieve via machine translation. Your content won’t be beautiful but if you follow these best practices, it will be comprehensible and effective.

Full article at http://www.business2community.com/tech-gadgets/translation-software-in-enterprise-0270669#RFFWYLYVp8XLmD1b.99

Veronica Lupascu wrote:
1. Make sure your source text is cleanly written and correct, free of idiomatic phrasings and generally machine translation-friendly. It’s also best if there’s a uniform tone throughout.
2. Your language vendor should establish upfront a dictionary of key terms. They should then have it reviewed and confirmed by all language stakeholders at your company.
3. Determine which approach — rules-based, statistical or hybrid — is best for your project. Statistical engines are appropriate if you have huge volumes of existing bilingual content. Otherwise, consider rules-based or a hybrid solution.
4. Invest in human post-editing for improved accuracy and quality. However, the editors should only change what is lexically essential to ensure understanding.
5. Have realistic expectations for the quality of the content you’re going to achieve via machine translation. Your content won’t be beautiful but if you follow these best practices, it will be comprehensible and effective.

Philippe Etienne wrote:

1. source text is written like the output of a machine
2. a dictionary is fed into the MT engine
3. TMs of old docs fed into the MT engine for a hybrid solution
4. if 1. and 2. are met, post-editing as per instructions (change what is lexically essential to ensure understanding) will be cheaper than human translation...
5. ...provided that you don't expect the outcome to smell, feel and look like human translation (content won’t be beautiful)